Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Evaluation and optimization of acoustical environments in eating establishments Razavi, Zohreh 2006

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2006-0632.pdf [ 12.95MB ]
Metadata
JSON: 831-1.0080713.json
JSON-LD: 831-1.0080713-ld.json
RDF/XML (Pretty): 831-1.0080713-rdf.xml
RDF/JSON: 831-1.0080713-rdf.json
Turtle: 831-1.0080713-turtle.txt
N-Triples: 831-1.0080713-rdf-ntriples.txt
Original Record: 831-1.0080713-source.json
Full Text
831-1.0080713-fulltext.txt
Citation
831-1.0080713.ris

Full Text

Evaluation and Optimization of Acoustical Environments in Eating Establishments by Zohreh Razavi B.Sc. Applied Physics, The University of Tehran, 1992 A THESIS SUBMITTED IN THE PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF M A S T E R OF APPLIED SCIENCE in The Faculty of Graduate Studies (Mechanical Engineering) The University of British Columbia October 2006 © Zohreh Razavi, 2006 ABSTRACT The acoustical environment (noise and reverberation) in eating establishments (EEs) is one of the most commonly overlooked design factors; and yet one of the most important. It can cause problems with verbal communication, especially for the hearing-impaired and second-language people. In extreme cases, too much noise can affect the health and safety of EE employees. Thus, to investigate acoustical environments in EEs, this research study was initiated. The objective was to evaluate the acoustical conditions to which customers and workers are exposed. It was to investigate how to optimize acoustical environments in EEs by way of room-acoustical design, acoustical treatment and noise control. Three EEs of three different types (restaurants, bistros and cafeterias) on and off the UBC campus were studied. To include hard-of-hearing people, two seniors' homes were also included. To evaluate the acoustical environment of each establishment, Reverberation Time and of noise levels in the unoccupied and occupied EEs were measured. RTs were longer than optimal in unoccupied and occupied EEs due to low absorption. Noise levels varied from 42.5 to 61.5 dBA in unoccupied situations, and from 55.3 to 74.5 dBA when occupied. The noise exposure of employees was also measured. Employees' daily noise exposure, on average, varied from 59.7 to 83.7 dBA, in compliance with the BC Worksafe regulations. Customer and employee questionnaires asking about the effect of the acoustical environment and factors that affect it, were developed, administered and analyzed. According to the responses, people visiting EEs tend to fall into two groups - one which visits for eating/drinking, talking and relaxing, the other for working/studying, business, celebration and relaxing - with different expectations of eating out. On average, they prefer to dine in a quiet environment with an appropriately low level of music. Sources of noise, such as people talking, moving, kitchen activities and equipment caused employees to suffer from fatigue, headache and tinnitus. Further analysis of measured occupied noise levels showed that customers used Casual to Raised voice levels for conversation, increasing their voice levels with increasing noise level (the Lombard Effect) at a rate of 0.7 dB/dB. Signal-to-noise level difference varied from-12 to 10 dBA. n Predicted voice levels with an existing "Lombard prediction" model which developed in parallel with this work was use to predict speech intelligibility and speech privacy at a typical seating position in an EE. Predictions were made of Speech Transmission Index and its components, Early Decay Time and A-weighted Speech to Noise level difference in one of the studied EEs using the CATT-Acoustic software. Different acoustical treatments (e.g. adding absorption, barriers, lowering the density and decreasing the receiver to talker distance) were evaluated. It was found that the most effective treatment was inserting absorptive barriers with the addition of absorptive ceilings, which led to good Speech Intelligibility and normal voice level. Reducing the distance between talkers and receivers is another solution, but this may not always be feasible. Table size, receive-talker relationship and the number of the people in groups limit this solution. i i i Table of Contents Abstract ii Table of contents iv List of Tables -- vi List of Figures 7 vii Chapter 1 Introduction 1 1.1 Acoustical Environment in EEs 1 1.2 Employee Noise Exposure— 2 1.3 Room acoustics and room acoustics prediction methods 3 1.4 Speech Intelligibility and Privacy 8 1.5 Noise-control measures 10 1.6 EEs and Open Plan Offices (OPO) 10 1.7 Acoustic problems in a U B C EE 12 Chapter 2 Literature Review 13 2.1 Eating Establishments (EEs) 13 2.2 Lombard effect 16 2.3 Research Objectives 18 Chapter 3 Physical Measurements— . 19 3.1 EEs characteristics 19 3.2 Physical measurements 20 3.3 Results—- 23 Chapter 4 Questionnaire Survey — 27 4.1 Methodology 27 4.2 Results 28 4.3 Customer questionnaire 28 4.4 Employee questionnaire 44 4.5 Customer questionnaire correlations 48 4.6 Employee questionnaire correlations 51 4.7 Discussion 53 Chapter 5 Advanced Noise Analysis - 55 iv Chapter 6 Prediction 64 6.1 CATT-Acoustic 64 6.2 "Lombard prediction" model 65 6.3 Method 68 6.4 EE configuration — 71 6.5 Results 74 6.6 Summary 80 Chapter 7 Conclusion 81 7.1 Research Contributions 81 7.2 Future Work 83 Bibliography 84 APPENDIX A - EE Photographs and Floorplans 87 APPENDIX B - Ethical Certificate 97 APPENDIX C - Consent Form 98 APPENDIX D - Employee questionnaire 100 APPENDIX E - Customer questionnaire 104 List of Tables Table 1. Main physical and acoustical characteristics of the EEs 25 Table 2. Employee job titles, shift lengths, measured average full-shift noise levels, Leq, and daily noise exposures, Lex 26 Table 3. ANSI S3.5 reference mean octave-band free-field sound-pressure levels at 1 m on-axis and sound-power levels for speech at five vocal efforts 59 Table 4. The optimal values defining the Lombard model for the EE studied 70 Table 5. The STI values along with their corresponding SI 71 Table 6. Materials used in the different prediction configurations, with the corresponding a and a 73 Table 7. Predicted acoustical values in different configurations with the talker-listener distance equal to 1 m, along with the changes due to occupancy: LO (12 customers and 4 talker) and HO (36 customers and 12 talkers); N = normal, R = raised, L = loud voice levels 76 Table 8. Predicted acoustical values in different configurations with the talker-listener distance equal to 0.5 m, along with the changes due to occupancy: LO (12 customers and 4 talker) and HO (36 customers and 12 talkers); N = normal, R = raised, L = loud voice levels 78 Table 9. Predicted acoustical values in different configurations with barriers 79 v i List of Figures Figure 1: Different possible paths from the source to the receiver in a room 4 Figure 2: Human speech directivity in the horizontal plane at 100, 400, lk, 4k and 10k Hz [11] — 7 Figure 3: Human speech directivity in the bilateral, symmetrical vertical plane at 100, 400, lk, 4k and 10k Hz [11] 7 Figure 4: Reverberation Time (R 7) measurement set up 21 Figure 5: Number of completed customer questionnaires from each EE 28 Figure 6: Percentage of female and male customers who completed questionnaires in the EEs: ( • ) Female, ( E3 ) Male, (W) Female averages, (0) Male averages ^ Figure 7: Number of customers in different age range categories Figure 8: Percentage of customers with First Language English (FLE) and Second Language English (SLE) in each EE: ( • ) FLE, ( H ) SLE, (0) FLE averages, (W) 30 SLE averages— — • Figure 9: Percentage of customers who were hard-of-hearing: ( • ) Yes, ( ) No, (W) Yes averages, (Id) No averages 3 ^  Figure 10: Preferences of customers for having meal in a quiet or noisy environment: ( • ) Quiet, (• ) Noisy, ( s ) Doesn't matter; averages: (M) Quiet, (l/l) Noisy, (H) 31 Doesn't matter Figure 11: Preferences of customers of having music with meal: ( • ) Yes, (• ) Doesn't matter, ( H ) No; averages: ( Y e s , (If) Doesn't matter (M)No 3 3 Figure 12: Average level of music that customers prefer to have with their meal 33 Figure 13: Preference of seating location in EEs because of noise concerns: ( • )Yes, ( • ) No; averages: (W) Yes, (0) No - — 3 4 Figure 14: Customers' reasons for visiting the EEs 35 Figure 15: Frequency of visiting EEs 36 Figure 16: How well the acoustical environment met customer expectations 36 Figure 17: Percentage of the customers who were annoyed by different sources of vii noise 38 Figure 18: Level of annoyance by different sources of noise in EEs 38 Figure 19: Percentage of the customers who experienced problems due to noise in EEs----- 39 Figure 20: Extent of problems experienced by customers due to noise in EEs 40 Figure 21: Extent of being overheard and of overhearing in EEs: ( ) Overhear, ( ) Overheard 41 Figure. 22: Percentage of employees who were bothered by different sources of noise: ( • ) SHs, ( • ) eateries 45 Figure 23: Extent of bother of employees by different sources of noise: ( - ) eateries, ( ) SHs 46 Figure 24: Percentage of employees who experienced different problems due to the noise: ( • ) eateries, ( • ) SHs 47 Figure 25: Extent of experiencing different problems due to the noise: ( ) eateries, ( ) SHs 47 Figure 26: Variation of L eq i Per (dBA) with number of customers in ten EEs, and the logarithmic regression lines through the data points. 56 Figure 27: Variation of pooled 7.eq,per with the number of customers with all EE data pooled, and the logarithmic regression line. 57 Figure 28: Variation of talker voice power level LM with number of customers in individual EEs. Also shown are logarithmic regression lines and reference voice levels. (C = casual, N = normal, R = raised, L = loud). — 60 Figure 29: Variation of SNA with Zeq,per m individual EEs, with linear regression lines. 61 Figure 30: Variation of SNA(dBA) with Leq, per(dBA) in all EEs, and the linear regression line. — 62 Figure 31: Variation of talker vocal power level LM with Zeq,Per in all EEs, and the 63 linear regression line and equation. — Figure 32: Proposed Lombard model used in this thesis. 67 Figure 33: Simplified geometric shape of BI showing the layout of the tables; spheres are receiver position at a typical seating position; the square box is the viii source (talker). — - — 69 Figure 34: Floor plan of computer model of EE BI showing the receiver positions (0-15) and talker position, A6. The dimensions are also given. The empty area is the customer traffic area without any seats. 72 Figure 35: Barriers in EE BI with the same distance from all the tables 73 Figure 36: Predicted octave-band RTs in the model EE when untreated. Tref is the measured RT. The other curves are the RTs predicted by diffuse field theory 74 Figure 37: Distribution of predicted EDT along seats 0 to 15 caused by a talker at seat 5. 79 ix Chapter 1 INTRODUCTION 1.1 Acoustical Environments in EEs Eating establishments (EEs - restaurants, bistros, cafeterias, dining rooms, etc.) are places where people go for many, different reasons. Although the main reason for customers visit is usually eating, other reasons, which may be equally important, include doing business, talking, reading, having fun, socializing, etc. The employees, on the other hand, go there to work and earn a living. The quality of the acoustical environment in an EE is an important factor that directly influences the health and well being of workers and customers, the quality of the eating experience and the reputation and commercial viability of the EE. Customers and employees are all different (young, old, normal-hearing, hard-of-hearing). Some people enjoy noisy environments, others hate them. Some activities may be enhanced by noise, others impaired. Most activities in an eating establishment involve talking and listening (i.e. verbal communication). Who has not dined in an EE that is so noisy or reverberant (echoey) that customers have to talk in a loud voice to communicate with friends and staff? Excessive noise will annoy customers who go to EEs with the expectation of being in a quiet area for reading and thinking, and who may not be able to concentrate. People who are particularly 'acoustically-challenged' (the elderly, hard-of-hearing and those using a second language) are seriously disadvantaged in noisy and reverberant environments [1], which may be effectively inaccessible to them. The acoustical environment is determined by the physical and acoustical characteristics of the EE, the strengths and locations of the noise sources, and the density, layout and demographic characteristics, etc. of the occupants. It is seldom considered in the design of eating establishments and few studies have considered EE acoustical environments and their impacts on workers and customers. We know that speech communication becomes more difficult in the presence of background noise. This phenomenon, called the 'Cocktail Party Effect', can be formulated as, "How do we recognize what one person is saying (the speech 1 signal to be heard) when others are speaking at the same time (the background noise)?" [2]. It is affected by the relative levels of the speech and the background noise, and the resulting signal-to-noise level difference. These depend on the seating density and occupancy, the acoustical characteristics, the sources of noise, and talker voice levels. When there is a high density of people in an EE, noise levels are high and customers raise their voices to be heard over the noise (the "Lombard effect") [3]. Customers raising their voices result in higher background-noise levels. A lower signal-to-noise level difference causes poorer speech intelligibility; however, it ensures higher speech privacy - customers do not feel that their conversation is being overheard at other tables. Customers and workers may have different needs for the acoustical environment. Depending on their personal characteristics and the objective of the visit, customers may have different expectations of the environment. In other words, a number of complex factors determine what constitutes the ideal or optimal acoustical environment in an EE. A non-optimal acoustical environment can negatively impact the work and dining experiences. Customers may be bothered by noise, and suffer detrimental psychological and health effects like stress and voice strain. Verbal communication between customers and staff may be difficult. Workers may suffer fatigue, stress, voice strain and hearing loss. Noisy conditions in eating areas can even reduce the appreciation for the food [4]. However, a noisy environment ensures speech-communication privacy and gives a 'dynamic ambiance'. 1.2 Employee Noise Exposure Too much noise can affect the health and safety of EE employees. Workers who work lengthy shifts with minimal breaks might be exposed to noise exposures that exceed levels permitted by applicable occupational-noise regulations. They may also experience fatigue, stress and voice problems, and could suffer more serious problems like hearing loss and time away from work. Inability to communicate well or hear warning signals from colleagues or equipment may put staff in danger. Continuous exposure to noise no louder than people shouting, for 8 hours a day, 5 days a week over a period of years can cause some degree of hearing loss [5]. This type of hearing loss is permanent. Workers who work lengthy shifts with minimal breaks might be exposed to noise that exceeds permitted by applicable occupational-noise regulations. 2 Noise is commonly measured in decibels (dB). The loudest sound pressure to which the human ear responds is ten million times greater than the softest, and the dB scale expresses this ratio by the use of logarithms. To obtain a measure which quantifies the sensitivity of the human hearing system to different sound frequencies, the A-weighted decibel (or dBA) was introduced [6]. People's hearing is most sensitive to sounds at mid frequencies. Thus, since the A-weighting amplifies mid-frequency sounds, the level of a sound in dBA is a good measure of the loudness of the sound. Leq, which is the equivalent continuous sound level of a noise energy-averaged over time, in dBA, quantifies the noise exposure of employees. Lex is the Leq, energy averaged over 8 hours, or the duration of exposure (e.g. a shift length) in dBA. Lex is useful, since it is comparable with occupational noise regulations for 8-hour shifts. Lex can simply be calculated from Leq as Lex = Leq +10 log t/8 [7], where t is the sampling period in hour. Based on guidelines for occupational health and safety issued by WorkSafeBC the daily noise exposure of workers should not exceed Lex = 85 dBA. Noise exposure of employees is related to room acoustics, which will be discussed more in the next section. 1.3 Room acoustics and room acoustics prediction methods Eating establishments are usually rooms and their acoustical environment is described by the theory of room acoustics. The main room-acoustic parameters which influence sound fields are: the room geometry, the acoustic properties of the room surfaces (absorption and type of reflection, which could be specular or diffuse depending on the surface properties), the sound source radiation characteristics (power, directivity, position), the receiver position and air absorption. In a room, there are a very large number of possible paths from the source to the receiver, involving various reflections from the room surfaces; the combination of all these paths determines how sound behaves in the room (see Fig. 1). To predict the acoustical conditions in rooms, different approaches can be taken. Diffuse field theory is one of the approaches, in which the theoretical assumption is that the reverberant sound field is perfectly diffuse. In fact, the diffuse field theory can be applicable in rooms with uniformly absorption distribution and diffusely reflecting surfaces and quasi-cubic shape. Although these assumptions are very restrictive, diffuse field theory is most widely applied to every type of room in practice, because of its simplicity. In this thesis, it was used to calculate 3 Figure 1: Different possible paths from the source to the receiver in a room. absorption coefficients from measured reverberation data. The room-acoustical parameters that can define the acoustical conditions in rooms are: steady-state levels (e.g. Background Noise Level; BNL) and reverberation. B N L in most rooms is usually the combined level of noise from various sources, such as outside noise, H V A C and other equipment; it varies with distance from the source and with the acoustical properties of the room. In EEs people are another source of noise. The B N L should not be so loud that it interferes with normal activity in the room, such as conversation or concentration. Conversely, it should not be so low that speech privacy isn't preserved. The background-noise criterion applicable to a given space depends on the normal activity there. Reverberation Time (RT) determines how reverberant the room is. It is the time taken for sound to drop 60 dB in level when a sound source stops abruptly. The equivalent total sound absorption area and the room volume determine the expected RT. The greater is the room volume and the more sound absorptive are the room surfaces, the lower is the RT. RT can be calculated from the Sabine equation [8]: RT = 0A6V A„ +4mV (1) 4 V is the volume of the room in m and A u = (a * S), where a is the average absorption of the surfaces and S is the total area of the room surfaces. The equivalent absorption, A u in m , is the number of square meters of a perfectly sound-absorptive material that has the same absorption. The 4mV represent the equivalent absorption in m corresponding to air absorption, with the air-absorption exponent in Np/m . A Neper (Np) is a reduction of energy of 1/e, and 1 N p = 8.7 dB [6]. Since the air-absorption contribution is proportional to the room volume, it is only significant in the case of large rooms, such as factories and concert halls, so it was ignored in this study. To calculate R T in the occupied situation, we need to add the absorption of the occupants, n A p , to A u . A p is the absorption of each person, taken to be 0.5 m 2 [9], and n is the number of people. 0.16V R T = — , s (2) The optimum R T for various types of rooms is different. A n average reverberation time of 0.4 to 0.8 s is about average for rooms where good speech intelligibility is desired for people with no hearing impairments. However, for hearing-impaired individuals, reverberation times should range from 0.2 to 0.5 s. It has been found that R T more negatively influences younger and older people than young adults [10]. Hard of hearing people need to converse in a place with lower R T compared with the normal hearing. To minimize R T is not necessarily a desirable goal, since it can result in severe attenuation of early reflections and ,consequently, speech intelligibility. The E D T (Early Decay Time) is the reverberation time, measured over the first 10 dB of the decay; it gives a more subjective evaluation of the reverberation time and it is more relevant to speech intelligibility than RT . The total sound pressure level Lp at a receiver, due to a source in rooms can be calculated with diffuse-field theory - the Sabine steady-state equation [8]: L p ( r ) = L w + 10 log e. + 4 4nrz R dB (3) 5 Lp is the sum of the direct-sound and reverberant sound contributions, Q and 4 respectively, Lw is the source sound power level. Closer to the source, sound levels depend on the properties of the source, and the direct field which is not affected by the room is dominant. Far from the source, sound levels are dominated by the reverberant sound field, which depends on the geometry and absorption properties of the room. At some distance from the source, which is called the " reverberation radius", the direct and reverberant sound pressures are equal; the reverberation radius indicates the ranges of distances where the direct and reverberant fields dominate and, thus, where the acoustical properties of the room do and do not affect the sound pressure. Lw is the sound power level of the source in dB. In the case of a human talker, values for various reference voice levels are defined in ANSI S3.5 as an average for male and female adults [11]. Q is the directivity of the source. For a source, which radiates the same energy in all directions, and located far from any surfaces (free field), Q can be considered as unity. The human voice directivity has a particular pattern (see Figs. 2 and 3); however in this study, Q is considered to be 2, which is the directivity of a human talker in 1kHz. R is the room constant in m 2 , and r is the distance between the receiver and the sound source in m. The room constant, R, changes with the total sound absorption and, therefore, the number of people inside the room. In the unoccupied room; R A, it 2 (4) =-,nr 1-4, / ( l - « ) 6 Figure 2: Human speech directivity in the horizontal plane at 100, 400, lk, 4k and 10k Hz [12]. Figure 3: Human speech directivity in the bilateral, symmetrical vertical plane at 100, 400, lk, 4k and 10k Hz [12]. 7 Another approach to predict room-acoustical conditions is based on ray-tracing and image-source methods [13]. The hybrid method combines the image source and ray-tracing methods together and, in short, it works as follows: rays are emitted from the point source and the surfaces hit by the rays are used as mirroring surfaces to calculate the position of the corresponding image sources. The program keeps track on all detected image sources in order to get one and only one contribution from each valid image source to the receiver [14]. As explained, in hybrid models, the image-source method is used to identify early reflections due to its accuracy in finding reflection paths, and ray tracing is used in later reflection calculation due to its efficiency of computational requirements. One of the hybrid acoustical predictions, CATT-Acoustic v8.0 [15], was used in this thesis. The Image-Source Model (ISM) is used to find early reflections and Randomized Tail-corrected Cone-tracing (RTC) for full-detailed calculation. The RTC combines features of both specular cone-tracing, standard ray tracing and ISM. Full-detailed calculations, calculate detailed quantitative analysis of echograms. 1.4 Speech Intelligibility and Privacy The acoustical design of rooms often involves optimizing speech-transmission performance, since speech communication is a basic function of public spaces such as EEs. Speech communication in EEs can be a difficult matter, since conversation at the tables becomes difficult due to high noise levels but, on the other hand, speech privacy is not ensured with low noise levels. The intelligibility of speech in rooms is influenced by both the signal-to-noise ratio, the difference in speech and noise levels, and the amount of reverberation in the room. The two effects are not independent; the reverberation can influence signal-to-noise ratio (S/N) by modifying the received sound levels of the speech and noise. The effects of S/N on the intelligibility of speech have been investigated thoroughly over many years, and measures such as Articulation Index (AI) have been developed to accurately indicate the expected effects of speech and noise levels. However, AI does not assess the effects of reverberation. Rooms add a complex sequence of reflected sounds that affects speech intelligibility and is often characterized simply by reverberation time. In fact, although some reflections degrade intelligibility, others can improve it. Our hearing system does not perceive individual early-arriving sound reflections as separate events; they are subjectively integrated with the direct sound and make the direct sound seem louder. Late-arriving reflections are not integrated with the direct sound and degrade 8 speech intelligibility by causing one speech sound to blur into the next. Typically, the boundary between early-and late-arriving reflections has been taken to be 50 ms after the arrival of the direct sound for speech sounds and, consequently, the Useful to Detrimental Energy Ratio, U50 has been used as an indicator for speech intelligibility with the effect of room reverberation [16]. This predictor divides the available acoustical energy (direct + reflected + noise) into a useful and detrimental part. The useful part consists of the direct energy from the talker, E d , and the early- arriving reflected energy, E e . The detrimental part consists of later arriving reflected energy, Ei and noise, E n . U 5 0 = lOlog Ed+Ee E, + E„ , dB (5) Increasing Ed, which is related to the sound power level of the talker, as well as increasing E e with more sound reflective surfaces close to the talker, could increase speech intelligibility. Decreasing E| and E„, which requires lower RT and B N L , respectively, could increase speech intelligibility. The Speech Transmission Index (STI) developed by Houtgast and Steeneken [17] is based on the concept of the modulation transfer functions. When the speech signal is transmitted through a room, the room reverberation and noise cause a decrease in the amplitude modulation of the signal. The modulation transfer function quantifies the changes in speech signal when it is transmitted through the room. These values are converted to effective signal-to-noise ratios and averaged into a single number from zero to one. STI assumes a value of 1.0 when all modulations are preserved and <1 when they are not. In this study, Speech Transmission Index (STI) was used as an index of quality for speech intelligibility and privacy. The Rapid Speech Transmission Index (RASTI) is a simplified version of STI, also developed by Houtgast and Steenken [18]. Like STI, RASTI ranges from 0 to 1, with a larger value indicating better speech intelligibility. However, RASTI is calculated based on only the 500 Hz and 2 kHz octave bands. Recently, AI has been replaced by the Speech Intelligibility Index (SH), as described in ANSI S3.5-1997. This is a little more complex to calculate than AI and includes reverberation, noise and distortion. Like AI it has a value between 0 and 1, but for the same conditions SII values are a little larger than AI values. An SII of 0.2 or lower is 9 considered acceptable speech privacy and values larger than 0.7 indicates good to excellent speech intelligibility. 1.5 Noise-control measures In general, to control noise in a room, an acceptable noise that does not interfere with the activities intended for the room, should be considered. In places intended for speech communication, quite low levels of noise will interfere with the quality of the communication; for example, levels higher than 35 dBA in classrooms or 50 dBA in EEs. The measured sound pressure level at the receiver position is related to how loud the sound source is, and will vary with the source-receiver distance and with the acoustical characteristics in the room. To control noise in one room, three factors need to be taken into account: noise sources, the path between source and receivers and the acoustical characteristics of the place. Noise sources could be "inanimate", such as equipment, H V A C or external noise, or could be "animate" such as people (conversation or movement). Inanimate noise sources are often constant and depend on the properties of the source. It is essential to control these sources by selecting equipment with adequately low noise output. Quieting the noise sources is more difficult after the design of the place. Noise due to peoples' conversations varies with their numbers, voice levels and the density of their distribution. Interaction between these sources of noise (inanimate and people) causes changes in the noise due to the people, since people talk louder in noisy places (Lombard Effect). The distances between noise sources and receivers give different perceptions of noise. At short distance, a casual voice level would be enough for good speech intelligibility; however, in EEs the size of the tables limits the distances. Noise control for good speech communication in EEs is not easy. The noise essentially consists of speech babble. It is not constant, because the customers do not always talk at the same time and their numbers fluctuate during the day. The noise from adjacent areas or outside, H V A C , kitchen activities and clinking sounds are the other sources of noise. 1.6 EEs and Open Plan Offices (OPO) Relevant to the acoustics of EEs is the acoustics of OPOs. In general, OPOs are large areas with large numbers of various types of office workers who are separated into work cells by barriers. There are common features between EEs and OPOs, since there are multiple sources 10 and listeners in both. Speech intelligibility is required within one work cell, speech privacy is required between work cells. The tables in EEs are the work cells in OPOs. In both EEs and OPOs it is desirable for the level of intruding speech to be low relative to the ambient noise, so that speech is less intelligible and speech privacy preserved. On the other hand, we need to have good speech intelligibility within one work cell in OPOs, or at one table in EEs. To have good speech intelligibility, the ratio of signal (voice level) to noise level should be adequate. It has been suggested that, for a face-to-face conversation, when lip reading, facial expression, and gestures contribute to intelligibility, a minimum signal-to-noise ratio of 5 or 6 dB is generally sufficient [19]. Such a ratio can be achieved either by decreasing the separation between listener and talker, which is limited by the size of the desks in OPOs or the tables in EEs, or by increasing vocal effort, where speech privacy won't be preserved. If the level of speech is high relative to ambient noise levels, then the speech will be quite intelligible. Although higher noise levels may better mask the unwanted speech sounds, the higher noise levels can become a source of annoyance and cause people to talk louder; hence they will not optimally improve speech privacy. So everything possible must be done to block the sounds from one work area so that they are not overheard in neighbouring work areas, but without the use of walls; in EEs the goal is to block speech coming from one table to the others. Electronic masking-noise systems are often added to OPOs to mask speech and other sounds from neighbouring work areas. However masking levels should be lower than 45 dBA, since levels greater than 45 dBA lead to occupants raising their voices, and annoyance [20]. Music in EEs can mask speech from adjacent tables as long as it isn't so loud that is annoying. In open-plan offices, to achieve acceptable speech privacy, the effects of various office design parameters have been investigated. The most important design parameters are: a) the sound absorption of the ceiling; b) the height of barriers between work cells; and c) the workstation plan area [21]. These parameters could be the most important factors in EE design, too, since a highly sound-absorptive ceiling could lower noise levels, as well as RT. Lowering the density of customers, equivalent to expansion of the workstations, could lower the background noise level, as well as preserving speech privacy. When the density of the customers is lower, they sit at larger distances from each other, so sound needs to travel further to reach to adjacent tables, making the sound pressure level lower and providing more speech privacy. Panels between tables would ensure more speech privacy, since they would block the direct sound. In OPOs speech is the most annoying source of sound, 11 according to a questionnaires studied [22]. Other parameters, such as speech effort, speaker orientation, speaker-to-listener distance and barrier attenuation were studied in OPOs. However, due to the different natures of the activities, the objectives of the occupants and the sources of noise, the results from OPOs are not directly applicable to EEs. A few papers in the literature about noise in EEs wi l l be discussed in the next chapter. However, before that it is important to explain a further motivation for this work. 1.7 Acoustic problems in a UBC EE At U B C , noise problems were experienced in the Sage Bistro (an upscale restaurant on the campus - referred to as R3 below) when some hard-of-hearing faculty members wanted to dine and converse in the main hall. They couldn't carry on their conversation due to the non-optimal acoustical conditions. That problem led to the idea of studying acoustical environments in E E ' s . 12 Chapter 2 LITERATURE REVIEW 2.1 Eating Establishments (EEs) Concern about controlling noise for good speech communication in eating establishments has led to a number of initiatives around the world. In 1993, Moulder [23] conducted a program to develop guidelines for restaurants and cafeterias, to provide quiet areas for hearing-impaired individuals that comprised 8.5% of the American population. The noise levels in the restaurants surveyed ranged from 55 to 68 dBA when averaged over a half-hour time interval during the busiest time of the day. The average reverberation time in the restaurants evaluated ranged from 0.36 to 0.95 s in unoccupied situations. They proposed the addition of sound-absorbing materials to rooms for the dual purpose of lowering background-noise levels and reverberation times. They suggested that the preferred way of adding absorption was the installation of an acoustical ceiling. In 1999, White [24] investigated the appropriate acoustic environment for an enjoyable meal, considering that communication with others around the same table is an essential part of the dining experience. The four dining spaces that were chosen for that study had a range of acoustical properties. Two different sets of questionnaires, one for diners and the other for listeners, who listened to a prerecorded conversation with background noise through headphones, were used. Correlations between physical measurement data (occupied noise level and RT) and questionnaire responses were found. There was a strong correlation between "enjoyment" and occupied noise level at the place for listeners, and less correlation for the diners. The diners' impressions about the physical properties of the rooms showed strong correlation between "hardness" and average RT values. White found a clear relationship between "ease of conversation", which was one of the questions on the questionnaires, and RASTI, as well as a strong correlation between "enjoyment" and a quieter level of noise. RASTI values at a distance of 0.8 m, which is a typical distance between a talker and the rest of the dining group, were calculated. Considering those responses from listener and diner questionnaires with correlations with RASTI of 0.75 or more, she found that, for RASTI values of 0.12 to 0.15, the acoustical 13 atmosphere was "indifferent", neither "annoying" nor "enjoyable". An "enjoyable" atmosphere was correlated to RASTI values of 0.3 to 0.45. From the speech-intelligibility curve developed from the intelligibility of short sentences with respect to RASTI values in auditoria, she found that low RASTI values of approximately 0.15 are at the bottom of the intelligibility scale (less than 10%), whereas RASTI values greater than 0.3 have very good intelligibility (greater than 95%). She found that, in a more absorptive dining room (average absorption coefficient of 0.7 versus 0.2), with the same number of people, the noise level is lower by nearly 15 dB. She compared RASTI in two different dining rooms (hard versus absorptive, with average absorption of 0.2 and 0.7, respectively) with different areas per person (1 m 2 , 1.5 m 2 , 2 m2) with two different assumptions that the number of talkers was half or one-third of the customers. She found that an acceptable RASTI of 0.3 could be achieved with an area per person 1.5 m 2 and an average absorption coefficient of 0.22 i f one-third of customers are talking at the same time. In 2000, Astolfi and Flippi [25] studied the optimal acoustical situation in pizzerias, as a compromise between speech intelligibility and privacy. Based on her study, since the major cause of noise in a dining space is the talking of customers, speech intelligibility and privacy targets could be achieved only by changing areas per person from 1 m 2 to 5 m 2 . She used a head and torso simulator as a speech source at a typical seating position 1 m in front of the receiver. With the source output corresponding to a normal vocal effort and an average noise level from measurements of 72.6 dBA, speech intelligibility was poor. However a raised voice level didn't preserve speech privacy with a distance of 1.5 m between the tables. She showed that speech privacy and intelligibility with a normal vocal effort were achieved with the seat area of 5 m per person. She used SII contours to show her results for different background noise levels, speech levels and customer densities. To take into account the increase of noise level with increasing number of customers, she used the Gardner [19] study which took the Lombard effect into consideration. Thus, she considered 6 dB increases in noise level per doubling of the number of customers. She suggested that the type of customer, and the rule of thumb that older adults are quieter than younger adults and families, should be taken into consideration. In 2002, Kang [26] studied the basic characteristics of speech intelligibility in dining spaces. A number of hypothetical dining spaces with different absorptions, sizes and seat densities were considered. He developed a computer model, R A D D , using the radiosity method. In the model all the boundaries are treated as diffusely reflective, which is the situation in many 14 dining spaces, as he suggested. He assumed that all the talkers have the same source power level, which was set at 0 dB, and the ambient noise at a listener was only from other diners. In other words, general background noise from other sources like the ventilation system wasn't considered in this study. He compared his computer model with diffuse field theory and concluded that although Sabine theory is useful in some dining spaces, it is more appropriate to use the computer model to analyze the speech intelligibility - in particular EDT and STI - when the sound field is not diffuse, due to unevenly distributed absorption or disproportionate room shape. Based on his results, increasing absorption is more efficient in improving speech intelligibility than enlarging the area per diner, and is relatively less expensive. He suggested that, i f the seat density is constant, the difference in intelligibility between different room sizes might not be significant. His results demonstrated that, when the ceiling height is increased, the speech intelligibility becomes lower i f the walls are acoustically hard, and can be higher when the walls are more absorbent. To investigate the effect of room shape, calculations were done with the length/width ratio of the room varied from 1 to 4. He found that, with a given floor area, the speech intelligibility can be improved by increasing the length/width ratio from 0.02 for a hard space to 0.07 for a space with sound-absorbent treatment. In 2004, Christie [27] studied the relationship between objective measures of RT, B N L and STI and their ability to predict a subjectively acceptable acoustical environment in bars, cafes and restaurants. The aim of the research was to see i f cafes, bars and restaurants are actually too loud or are acceptable acoustic environments for their occupants. They found that the establishments were too loud, or undesirable, from an objective point-of-view due to the excessive background-noise level in 60% of them, which had an average of 57.3 dBA in the bars, 65.0 dBA in the restaurants and 57.7 dBA in cafes. However, occupants adapted to the acoustic environment of the establishments by raising their voice levels. The authors found that the response to the question as to whether noise impairing the conversation was negatively correlated to STI values. They proposed that this negative correlation could suggest that people are most likely talking in a louder voice than as assumed in the STI calculations. They used predefined speech levels from ANSI 53.5 (1997) for STI calculations. They suggested that the ANSI speech levels are not, in fact, representative of normal, raised, loud and shouting levels of speech in these particular environments, and that they 15 needed to consider higher levels of speech as standards for different levels in these areas. However they didn't propose any new values. The Lombard effect wasn't considered in the Moulder [23], Kang [26] and Christie [27] papers. However, Astolfi et al. [25] addressed the Lombard effect, as mentioned, and White [24] noted this effect in EEs; though it wasn't considered in the predictions. 2.2 Lombard effect When people are speaking in an enclosure, the noise level increases as the number of people increases. This growth in noise level gradually interferes with dialogue, and lowers the speech intelligibility. To preserve the speech intelligibility, people have to raise their voices. The people keep on raising their voices, resulting in a rapid increase in the noise level; this positive feedback would continue until human limits stop the increase of voice level. This is the so-called "Lombard effect", first discussed by Lombard in 1911 [28]. It is the "adaptations of speech to overcome the detrimental effects of noise in noisy places". The Lombard effect is a nonlinear distortion which depends on the speaker voice level, the background-noise level and the type of noise. There are a number of papers in the literature about this effect. In 1954, Korn [29] found that noise levels below 45 dB do not seriously influence speech power. He suggested that higher noise levels - over 55 dB - influence speech power, and result in a 0.38 dB increase in speech level for every 1 dB increase in the noise level (a Lombard slope of 0.38 dB/dB). His subjects were fifty men and women between ages 20 and 60. White noise was fed to earphones worn by the speaker, with a large gap between the speakers' ears and the earpieces so that, in the absence of noise, the speaker was virtually in the same acoustical conditions as without headphones. The measurements were made by asking them to read a few selected sentences in noise levels, which were increased in steps of 10 dB from 40 to 90 dB and then decreased in the same manner. The overall frequency response of their speech was measured from 50 Hz to 10 kHz using a microphone close to the speaker's lips. In 1957, Pickett [30] used an anechoic chamber to study the Lombard effect in free-field acoustic conditions. Listeners weren't able to see the talkers' lips. The listeners checked the words that they heard to verify the sentence correctness; the vocal efforts of talkers in different noise levels were monitored by tape recorder and expressed as the rms speech level in a free field at one meter from the talkers' lips. He found an increase of 8 dB 16 in mean vocal effort with an increase in noise level of about 8 dB (a Lombard slope of 1 dB/dB). In 1971, Gardner [19] studied the magnitude and rate of vocal output changes for group situations. He measured that, in an auditorium, for each doubling of the number of loudspeakers in use, the total output level increased by 3 dB. However, for each doubling of the number of people, the total vocal output increased by 6 dB. That rate of change in total vocal output level with doubling the number of people occured once the audience exceeded 12-15 people, on the assumption that one-third of them talk at the same time. In 1997, Tang, Chan and Chan [31] studied the Lombard effect in the staff canteen at the University of Hong Kong. He showed that occupants raised their voices when B N L exceeded 69 dBA. This effect might not be seen when the number of occupants is less than 50, due to the large difference in individual voice levels and the granularity of where people sit in the place. They also developed a prediction model based on the assumption that, in noise, talkers raise their voice to maintain constant signal-noise level difference. It predicted measured noise levels well. In 2004, Whitlock [32] asked 18 children to read from a book in an anechoic chamber while different masking noises were fed to their ears. He found that the overall response was nonlinear, but indicated that the results could be approximated as linear above the base voice level of 53.4 dB with a slope of 0.22 dB per dB of masking noise. Bradley [33] found this rate to be 0.82 dB/dB through noise measurements in 41 classrooms. In 2005, Sutherland and Lubman [34] developed a formula for the Lombard effect based on the data published by Pearson, Bennett and Fidell [35]. It assumed that voice levels begin to rise above a noise level of 35 dBA. These included data for teachers' voice levels in 18 classrooms, where the Lombard rate was essentially 1 dB increase in the teacher's voice level for 1 dB increase in the noise level. They modeled the Lombard effect empirically, by assuming that the teacher's voice level at 1 m would be equal to the energy sum of the teacher's voice level in quiet at 1 m, L0 (about 57 dBA) added, on an energy basis, to the sum of a background noise level, Ln, and a Lombard constant, K(L). A Lombard constant of 22 to 26 dB was considered, to accommodate other Lombard rates, R(L), than 1 dB/dB. The speaker voice level in noise, Ls (n), was formulated as: Ls (n) = \Q log [10A (Lo/\0) + 10A{{ La [1- R(L)} + R(L) [L„ + K(L)]}/10}, dB (6) 17 2.3 Research Objectives The objectives of this study were to evaluate the acoustical environments, and thus the acoustical conditions to which customers and workers are exposed, in eating establishments. It was to investigate speech and noise levels, and the Lombard effect, in EEs. It was also to investigate how to optimize acoustical environments in EE's for a particular type of customer, by r way of room design, acoustical treatment and noise control. These objectives were achieved by performing acoustical measurements, administering questionnaires, and developing and applying a prediction model. The specific objectives of this study were to: > Identify EEs of different types to include in the study; > Characterize and evaluate the quality of the acoustical environments in the EEs by way of physical measurements and existing acceptability criteria; > Develop two different versions of questionnaires (customers and employees); > Characterize and evaluate the perceived quality of the acoustical environments in the EEs by way of customer and employee questionnaires; > Investigate how the perceived quality of the acoustical environment varies between workers and customers, and how respondent-related modifying factors affect it; > Investigate the relationship between respondent perceptions of quality and the acoustical characteristics of the EE; > Identify the optimum EE acoustical characteristics and how these vary between workers and customers, as well as with respondent-related modifying factors and EE acoustical characteristics; > Analyze noise in EEs to find typical speech and noise levels, and the Lombard constant and rate, using "Lombard prediction" to find noise levels with different occupancies, and apply them in predictions; > Use a room prediction model to investigate how to modify and improve the acoustical environments in EEs. 18 Chapter 3 PHYSICAL MEASUREMENTS 3.1 EE characteristics Six typical eating establishments on the U B C campus and four off the campus, of four different types - restaurants (designated R), bistros (B), cafeterias (C) and seniors'-home dining rooms (S) - were studied. These were chosen on the basis of convenient location, access through existing contacts, their physical characteristics and customer demographics. Appendix A shows photographs and floor-plans of the EEs. Eating establishments on the campus included the Student Union Building cafeteria, which served university student customers. This EE has different areas with very different acoustical environments. Two different areas, C l and C2, were chosen. In both places the floor was carpeted. In C l the ceiling was covered with acoustic tiles, whereas C2 had an unfinished ceiling with exposed beams. In C l furnishings were hard with thin vinyl chairs and wooden tables, whereas in C2 there were wooden tables, with upholstered wooden chairs. Both places were large, rectangular-shaped spaces in a multi-dining area, with the different areas separated from one another by half-height brick or plasterboard walls. One of the external walls mostly contained windows, and the rest of the walls were painted drywall. There was no music in either of the EEs, and the main source of noise was people (talking and moving). Restaurant R3, which serves university faculty customers, had more acoustical treatment than the other EEs on the campus. In the main eating area where the measurements made, the ceiling had some regions with acoustic tiles concealed behind large white drapes. The walls contained large windows or were painted drywall. The floor was wooden in the main area; however there were two smaller mezzanine areas with carpeted floor. The tables were wooden with thick tablecloths; the chairs were wooden with padded seats. Music added 3 dBA to the background-noise level when it was on. Bistro B l , which serves university student and faculty customers, has large windows, and a hard floor and ceiling without any acoustical treatment. The tables are metal with thick tablecloths; the chairs are metal with wooden seats. Music was not played on the day of measurement. Bistro B2 serves university student and faculty customers, and has the same construction and furnishing as B l . Bistro B3 has indoor (B3/in) and 19 outdoor (B3/out) areas with very different acoustical environments. The indoor area was surrounded by large windows, and had wooden chairs and tables, with an open kitchen area and a loud level of music. The open kitchen area, with a noisy fan and loud music, creates a really noisy ambiance. Music added 6 dBA to the background-noise level when the EE was unoccupied. The outdoor area is not exposed to the noisy fan of the kitchen, but music is played from loudspeakers. The furnishings in the outdoor area were typical patio vinyl chairs and tables. t. Restaurants off campus, R l and R2, with very different acoustical conditions, were enlisted, to involve older, more non-academic customers. In R l , the walls are predominately windows; the ceiling and floor are hard. The furnishings are also hard, with marble-topped tables and wooden chairs. The EE is located on a busy road, and next to a bus stop. In the unoccupied situation, music added 9 dBA to the level of the background noise. R2 has more acoustical treatment, with carpeted floor and some suspended drapes on the ceiling. The furnishings were wooden tables, and chairs with vinyl seat pads. There was no music on the sampling day. Finally, dining rooms in two seniors' homes, SI and S2, were enlisted to involve more elderly, hearing-impaired customers. Both dining rooms had carpeted floors. The ceiling in S2 was covered with acoustic tiles; in SI it was of painted drywall. In S2, tables were wooden with upholstered chairs. The furnishings in SI were wooden chairs and tables. There was no music in either seniors' home. 3.2 P h y s i c a l measurements In each EE, physical acoustical measurements were made to characterize the acoustical environment. Following are details of the measurements: a. Reverberation Time (RT): Reverberation Time is the time in seconds that it takes for a sound to decay by 60 dB. RTs were measured in the unoccupied situation; because high sound levels must be generated to measure reverberation times, it is not feasible to make these measurements with the customers present. Measurements of impulse response between a source and varied receiver positions were made using the M L S S A system, involving generating noise bursts fed to an amplifier and an omni-directional loudspeaker system, as shown in Fig.4. 20 Figure 4: Reverberation Time (RT) measurement set up, including MLSSA system, amplifier and omni-directional loudspeaker. Octave-band RTs were measured from 125 to 8000 Hz at six different locations in each EE in the unoccupied situation. /?rav,unoccupied was calculated by averaging over all locations and bands. m^id,unoccupied was calculated by averaging the 500, 1000 and 2000 Hz octave-band frequencies most related to speech intelligibility. The absorption of the occupants can decrease the RT. Thus, diffuse-field theory was used to account for the absorption of the EE occupants, in order to derive iJT'mid.occupied values. The average occupancy (number of customers) in each EE was considered, and the average absorption per person was assumed to be 0.5 m2. For acceptable verbal communication, especially for acoustically-challenged listeners, mid-frequency RTs should not exceed about 0.5 s when the EE is occupied; b. Noise levels: Background-noise levels (BNL) - total A-weighted noise levels in the 125 to 8000 Hz octave bands - were measured when the EE was unoccupied. A RION NA29E sound level meter was also used to identify the significant noise sources (e.g. ventilation outlets, kitchen equipment) and their noise levels. The 20s, A-weighted, equivalent sound pressure level was calculated with the help of Excel program in octave-band frequency. Larson-Davis 700 noise dosimeters were also used, to measure total, A-weighted noise levels in the EE 21 due to customer and staff activity and noise sources when the EE was occupied, to determine typical noise levels to which customers were exposed. They were located on the ceiling. Their microphones were hung about 50 cm below the ceiling. The dosimeters recorded average A -weighted sound-pressure level (Leq) every 60 s during operation hours of the EEs. Noise dosimeters were measured noise due to the customers and staffs activities. The data then was imported into MS Excel and decibel averaging was use to determine the average sound pressure level, Leq, over the operation hours. To ensure speech privacy, ambient noise should be in optimal situation and not lower than customer's speech level. Background noise levels recommended for bars, cafes and restaurants are 45-50 dBA, 45-50 dBA and 35-50 dBA, respectively. In general, An acceptable level of noise for verbal communication can be considered to be 35 dBA for acoustically-challenged people, and 45 dBA for others; c. Employee noise exposures: Noise dosimeters were also employed to monitor full-shift noise exposures of employees in different job categories who volunteered to be involved. Total, A-weighted noise levels in the workers' hearing zones were monitored using Larson-Davis 700 dosimeters, using standard sampling parameters (criterion level = 85 dBA, threshold level < 80 dBA, exchange rate = 3 dBA, time constant = "slow"). The dosimeters were attached to the workers' waist using a belt-mounted pouch, and the microphones were clipped onto the workers' lapel. The device weighs less than 1 kg, so it was light and it didn't interfere with workers' regular tasks. Noise dosimeter is a standard instrument for measuring the noise dose in the work place. The instrument does not record or transmit conversation, and therefore any dialogue or speech cannot be retrieved from the data collected. So workers can treat their workdays as any normal workday when they are wearing one of the dosimeters. The dosimeter recorded equivalent-continuous ("average") noise levels every 60 s. The data were downloaded from the dosimeters to computer, imported into MS Excel and decibel-averaged, to determine the full-shift equivalent continuous ('average') sound levels, Leq. Daily noise exposures (Z e x) over the monitoring periods were calculated from the full-shift Z e q ' s and daily shift lengths, and compared with the maximum acceptable level (Z,ex,wcB = 85 dBA) specified by the WorkSafeBC occupational-noise regulations [36]. 22 3.3 Results Table 1 shows the main characteristics of the EEs. The volume of the EEs varied from 176 m 3 in RI to 1176 m 3 in SI, with the surface area varying from 202 to 876 m 2 and the floor area from 30 to 294 m . The volume-to-surface-area ratio varied from 0.84 m in R2 to 1.34 m in SI. The number of seats varied from 40 in RI to 126,.in R3. The seat density (the number of seats divided by the floor area) varied from 0.4 1/m2 in SI to 1.3 1/m2 in RI . The average occupancy (based on customer estimations of the number of the people in the EE, averaged over the measurement period, and then divided by the total number of seats) varied from 29% in B2 to 95% in C2. -KTaVjunoccupied varied from 0.5 s in R2 and S2 to 1.2 s in B l . ^Trnj^unoccupied varied from 0.5 to 1.5 s. ivTmJd,occupied varied from 0.45 s in R2 to 1.41 s in B l . Table 1 also shows the average surface-absorption coefficients ot a v g calculated from ./?/^unoccupied and the EE surface areas using diffuse field theory. Values vary from 0.16 in B3 to 0.35 in R2. Due to the low absorption coefficients, values of/vTmid^ccupied in B l , B2, B3, RI , R3 and SI were higher than the optimal value of 0.5 s. In C l , R2 and S2, absorption coefficients were higher than average and, consequently, ^rmid.occupied was shorter than 0.5 s. SI also has high absorption, but this is offset by a high volume-to-surface-area ratio giving a lower RT. BNL varied from 42.9 dBA in C l to 61.5 dBA in B3 in the unoccupied situations, and from 55.3 dBA in S2 to 74.5 dBA in B3 when occupied. Customers in B3 need to talk louder than a raised voice level (72 dBA on average for male and female talkers) to overcome the noise. The high BNL in the unoccupied situation in B3 was due to the noisy ventilation system in the kitchen, which affects the eating area. The increase in BNL varied from 0.1 dBA with 37 customers in S2 to 26.9 dBA with 108 customers in C l . Due to the extensive acoustical treatment in S2, BNL didn't change between the unoccupied and occupied situations. The same average occupancy in B l and RI caused BNL in both places to increase by the same amount when occupied. The absorption coefficients in B l and R3 are comparable; however, due to the larger surface area and, consequently, more absorption in R3, BNL in the occupied situation increased less, even though the occupancy was higher. Table 2 summarizes the employee noise-monitoring results. It shows the job title, shift lengths, measured full-shift Lcq and corresponding daily noise exposures Lcx. Full-shift equivalent continuous levels Z e q varied from 61.7 to 83.7 23 dBA. Employee daily noise exposures Lex varied from 59.7 to 83.7 dBA, with an average of 75.3 dBA. The server at the SI seniors' home who worked the morning shift had the lowest exposure; the cook and cashier at B3 had the highest exposures. Considering the ±2 dBA uncertainty associated with field measurements [37], the daily noise exposure of the employees at B3 may be excessive. The noise exposures of the employees in the EEs were, on average, 7 dBA higher than in the seniors' homes, due to the different customer demographics. 24 Table 1. Main physical and acoustical characteristics of the EEs. B3 BI B2 R3 C l C2 R l R2 S2 SI Volume (m3) 333 692 j 384 960 619 412 176 180 297 1176 Surface area (m2) 393 599 314 812 584 424 202 215 315 876 Floor area (m2) 85 69.5 47 240 221 147 30 65 99 294 Volume/Surface area (m) 0.85 1.16 1.22 1.18 1.06 0.97 0.87 0.84 0.94 1.34 Number of seats 70 72 46 126 120 100 40 54 56 106 Seat density (1/m2) 0.8 1.0 1.0 0.5 0.5 0.7 1.3 0.8 0.6 0.4 Average occupancy (%) 86 33 29 40 90 95 34 56 66 38 RTay (s), unoccupied 0.8 1.2 0.9 0.8 0.6 0.9 0.8 0.5 0.5 0.7 RTmi6 (S), unoccupied 0.9 1.5 1.2 0.8 0.5 1.0 0.9 0.5 0.5 0.8 ^ T m i d (s), occupied 0.6 1.0 0.8 0.7 0.4 0.5 0.7 0.4 0.4 0.7 ct a v 0.16 0.23 0.22 0.24 0.34 0.19 0.18 0.35 0.30 0.29 BNL (dBA), unoccupied 61.5 51.5 60.0 54.7 42.9 55.1 53.3 57.4 55.2 49.1 BNL (dBA), occupied 74.5 67.1 69.0 69.2 69.8 70.4 70.3 70.4 55.3 59.4 25 Table 2. Employee job titles, shift lengths, measured average full-shift noise levels, L, and daily noise exposures, Lex. Eating Establishment Job title Shift length (hours) L e q (dBA) Lex (dBA) B3 Cook 7.75 83.6 83.4 Cashier 4.0 86.7 83.7 EEs on the U B C Cl/2 Janitor 3.5 83.0 79.4 B l Cook 6.5 76.0 75.0 Cashier 4.5 73.5 71.0 campus B2 Cook 7.5 80.8 80.7 Cashier 6.5 70.5 69.5 R3 Cook 6.5 79.9 78.9 Waitress 4.0 80.0 77.0 EEs RI Cook 6.0 76.4 75.1 off the Waitress 7.0 76.5 76.5 UBC R2 Cook 3.5 75.4 72.4 campus Waiter 4.0 75.8 72.8 Seniors' S2 Server 5.5 81.4 79.0 homes SI Server 7.0 71.4 70.9 Server 5.0 61.7 59.7 Min 3.5 61.7 59.7 Max 7.5 86.7 83.7 Average 5.4 77.0 75.3 Std 1.0 6.0 6.0 26 Chapter 4 QUESTIONNAIRE SURVEY 4.1 Methodology Questionnaires were developed to determine EE occupants' perceptions of the acoustical environments, and factors that might affect them. Two different versions of the questionnaire (for customers and employees) were developed. The questionnaires are presented in Appendices C and D. They comprise two major parts - respondent-related questions and acoustical evaluation of the eating establishment. The respondent-related questions solicit information about the respondent (age, sex, English-language ability, hearing status, the number of people in their party, the objective of the visit to the EE, music and noise preferences). The acoustical-evaluation questions ask respondents about the perceived quality of the acoustical environment of the EE with respect to the optimal environment, about annoyance due to sources of noise (e.g. the ventilation system, people talking or moving, music), about detrimental psychological and health effects (e.g. fatigue, stress, tinnitus, voice strain), and about verbal communication and privacy. A l l questionnaires included the floor-plan of the EE; customers and workers were asked to indicate their seating location and working area(s), respectively. At the end of the questionnaire, they were asked for their comments on acoustical improvements to the establishment or on any other important issues they wished to comment on. For questionnaire administration, customers were recruited via posters placed in each establishment, or were asked by the EE employees to participate in the study. Questionnaires were taken by customers from a designated place on the cashier's desk, or were given to the customers by the EE employees. A l l questionnaires included relevant contact information to allow subjects to pose questions to the researchers. With the help of the EE manager, employees were asked to complete the questionnaire on the study day. Completed customer and employee questionnaires were checked for completeness. On average, seventeen completed customer questionnaires and three employee questionnaires were analyzed in each EE. Questionnaire responses were coded numerically for use in correlation analysis. 27 4.2 Results In this section the questionnaire responses are discussed. Results are generally presented individually for each EE; however, in many cases average results for the EEs, which are not seniors' homes - these will be referred to as "eateries" (thus, Avg-eateries) - and for the seniors' homes (Avg-SHs) are also presented. In bar graphs, averages are shown in lighter colors or in line graphs with dots. Following this, the results of correlating different responses, and correlating the questionnaire responses with measured data, are presented. 4.3 Customer questionnaire The number of completed questionnaires returned by customers in each EE is shown in Fig. 5. The number of completed customer questionnaires varied from 6 in S2 to 26 in the C l , with an average of 17. 30 Figure 5: Number of completed customer questionnaires from each EE. 28 Customers were asked about certain personal characteristics, such as sex, age, English-language ability, environment preference and hearing status. As shown in Fig. 6, more than 60% of the customers in all of the eateries were female, except in BI with only 30% female customers. In seniors' homes, there were approximately equal numbers of males and females. As can be seen in Fig. 7, the majority of the customers who frequent the EEs were in younger age groups, between 20 and 30 years old. There was a high number of customers in theB2, B3 and Cl /2 with ages of 20 to 25 years old, and in BI between 25 and 30 years old. Many R3 customers were around 50 years old. No customers older than 75 years old completed questionnaires in the eateries; however, the average age of the participants in seniors' homes was greater than 75 years. 29 30 Figure 9: Percentage of customers who were hard-of-hearing: ( • ) Yes, ( • ) No, Figure 10: Preferences of customers for having meal in a quiet or noisy environment: ( • ) Quiet, ( • ) Noisy, ( Q ) Doesn't matter; averages: (E^) Quiet, ( 1 / 1 ) Noisy, ( H ) Doesn't matter. Customers were asked about their language ability, to see whether English is their first language or not. As shown in Fig. 8, on average more than 70% of customers in the EEs had First Language English, with a minimum of 30% in C2, and a maximum of 95% in R3; on average, 90% of the respondents in the seniors' homes had First Language English. No Second Language English residents in S1 participated in this study. As shown in Fig. 9, there was an insignificant number of hard-of-hearing respondents in the eateries (the highest proportion was 20% in R3); in the seniors' homes, on average, 45% of the respondents were hearing-impaired. However, on average they reported that the severity of their hearing-impairment wasn't greater than "moderate". Customers were asked about their preference of having their meal in a quiet or noisy environment. As shown in Fig. 10, on average more than 58% of customers in EEs prefer to dine in a quiet environment, for more than 38% of the customers it doesn't matter, and just 4% of respondents preferred to have their meals in a noisy environment. A quiet environment was especially preferred in R l , R3 and the seniors' homes. In seniors' homes no-one preferred to have their meal in a noisy environment. Customers were asked their preferences of having music with the meal during their visits to the EEs. As shown in Fig. 11, customers in B3 mainly preferred having music with their meal. In C l , R2, R3 and SI customers mainly preferred not having music with their meal. On average, respondents in the eateries preferred music more than in seniors' homes. 32 33 Figure 13: Preference of seating location in EEs because of noise concerns: ( • ) Yes, ( • ) No; averages: (63) Yes, (fcl)No. As shown in Fig. 12, even customers in B3 who preferred to have music with their meal, don't want it to be played at louder than "moderate" level. On average, customers in R3 who preferred music with their meal, preferred it to be played at lower than "low" level. Averages for EEs are shown with hatched barrs, and indicated that all patrons preferred music not to be louder than at "moderate" level. Customers were asked to indicate whether they preferred a particular seat location because of noise concerns. As shown in Fig. 13, generally they did not, except in B3 (presumably because of the noisy H V A C system and loud music from loudspeakers), and in S2 (presumably because of the noisy, open kitchen area). Customers were asked to indicate the objective(s) of their visit to the EEs, and how well the acoustical environment met their expectations of an ideal environment relative to the visit objectives. As shown in Fig. 14, and as expected, the most common reason for people to frequent these EEs was eating/drinking (less in R3), followed by talking (especially in C l and C2) and relaxing (less in R3). Dining out with the objective of having a business meeting, and of studying/working, were the reasons for coming to the EE for some customers (especially in C l and R3, less in B3). In seniors' homes, eating/drinking was the only reason for the seniors to visit the dining room. 34 100 90 80 70 3" 60 a O) | 50 Q) O L_ (0 a. 40 30 41 20 10 I Eating/Drinking B Talking EBus iness E3 Studying/workin j g 0 Celebration • Relaxing Figure 14: Customers' reasons for visiting the EEs. Customers were asked to indicate how frequently they visit the EEs. As shown in Fig. 16, in eateries, on average, there wasn't any significant difference in the percentage of the people who visit EEs with different frequencies. However in C l and C2, more than 70% of the customers visited "every day"; in BI this percentage was more than 60%. Customers who participated in this study visit B3/out "sometimes" or more. However, more than 50% of the customers in R3 visit there not more than "sometimes". In seniors' homes more than 80%) of the . residents visit the eating area "every day". 35 100 90 Every day • Often B Somet imes • First time a> cn ra c d> u u 0) Q. & <?> EEs Figure 15: Frequency of visiting EEs. Very Well Well Averag Poorlyj Very Poorw — Eating/Drinking -Talking — « — Business — * - — Studying/working| Celebration Relaxing Figure 16: How well the acoustical environment met customer expectations. 36 Fig. 16 shows the customers' reports of how well the acoustical environment of the EE met their expectations with respect to their visit objective(s). Triangles are shown the average responses of the customers who visited the place for each objective. On average, expectations were better met for those customers who visited EEs for business, studying/working and celebrations, and less well met for those who visited EEs for eating/drinking, talking and relaxing. For talking, R l met expectations the most, R2 and B3/out the least. Customers in R l reported visiting it more often than customers in R2, and that could explain the higher degree of met expectations. Customers who visited the B3 and were seated on the patio with the objective of business found the place to meet their expectation "well". Customers who frequented R l with the objective of business and of studying found it to meet their expectations "very well". On average, the expectations of customers who visited EEs for relaxation were met not better than "average"; even in B3 (inside and outside) and R3 their expectations for relaxing were met "poorly". On average, expectations were better met for residents in S2, and less well met in SI. On average, customers in eateries found their expectations met better than in seniors' homes for eating/drinking, talking, relaxing and business. That may partly explain why, in seniors' homes, residents only visit the dining room for eating/drinking! In the acoustical-evaluation part of the questionnaire, customers were asked to indicate which sources of noise were most annoying, and how much they were bothered by those sources during their visit. Fig. 17 shows the percentage of the customers who reported annoyance due to the different sources of noise. On average, in all eateries except in R3, 100% of the customers were annoyed by all sources of noise. However, in seniors' homes, 60% of the respondents reported annoyance by people speaking and more than 50%> of them reported clinking to be as annoying as adjacent-table conversation. Noise due to kitchen activities wasn't a significant source of annoyance in seniors' homes. 37 • Peop le speak a Peop le move Q Adjacent conversat ion H Cel l phones • H V A C 0 K i t chen activities = K i tchen equipment E Cl ink ing BOutside no ise E M u s i c m Reverberation Figure 17: Percentage of the customers who were annoyed by different sources of noise Very Low EEs ^ ^ " - P e o p l e speak ing "" P e o p l e mov ing - 0 — Adjacent conversa t ion - X — C e l l p h o n e s H V A C • K i t chen activit ies Kitchen equipment -e—Cl ink ing < Outside noise Music — — — • Rov/orhoratinn Figure 18: Level of annoyance by different sources of noise in EEs. 38 100 90 80 * 7 0 aT 60 O ) 50 40 30 20 10 0 c U L . Oi C L a* EEs V" T i red vo ice H e a d a c h e Fat igue Diff iculty in hear ing conversa t ion Diff iculty talking with wai ter /wai ters l B r o k e n concentrat ion I R e d u c e d enjoyment Figure 19: Percentage of the customers who experienced problems due to noise in EEs. As shown in Fig. 18, on average, customers were most annoyed by people speaking and adjacent-table conversation, and least annoyed by outside noise. Annoyance was generally "very low" or "low". On average, residents in seniors' homes reported more annoyance from sources of noise than did customers in eateries. In C l and C2, R2 and SI, customers found people speaking to be the most annoying source of noise. In C l and C2, adjacent-table conversation was reported to be as annoying as people speaking. However, in B3 (inside and outside) music was rated as the most annoying noise source. In RI and S2, clinking sounds were found to be the most annoying. In RI , reverberation was reported to be as annoying as kitchen equipment. In B2, kitchen activities, kitchen equipment and clinking sounds were reported to cause the same annoyance for customers. In B l , reverberation, people speaking and adjacent-table conversation were the most annoying sources of noise. 39 Very Low! 1* Tired voice H e a d a c h e Fat igue Difficulty in hearing conversat ion Difficulty Ta lk ing with waiter/waiters • B roken concenta t ion • R e d u c e d en joyments Figure 20: Extent of problems experienced by customers due to noise in EEs. Questions were asked about problems customers experienced in EEs due to the noise while they were dining out. Fig. 19 shows the percentage of the respondents who experienced different problems due to the noise in EEs. The problems have been divided into three types: "health- related" (fatigue, headache, tired voice - marked Black); "linguistic" (difficulty talking with waiter/waitress and difficulty hearing conversation at the table - light grey); and "psychological" (reduced enjoyment and broken concentration - dark grey). On average, shown with hatches, more than 90% of the customers in the eateries, except in R3, experienced all of the problems. In S2, 70%> of the residents experienced "linguistic" problems. In SI, 60% of the residents experienced difficulty hearing conversation at the table, which caused reduced enjoyment. On average, customers in S2 experienced all problems more than those in SI; the average age of the participants in S2 was older than in SI, and that could be the cause. The extent of the problems customers experienced in EEs is shown in Fig. 21. On average, all problems were experienced more in B3/inside than in the other EEs, especially "psychological" problems. However, "psychological" and "linguistic" problems were experienced most in C l and C2. 40 Very Well Well Figure 21: Extent of being overheard and of overhearing in EEs: ( ) Overhear, ( ) Overheard. Customers were asked to indicate to what extent they could easily overhear adjacent-table conversation, or felt they could be overheard. Customers who indicated that they could overhear others usually reported that they were also overheard, as shown in Fig. 21. In B3/in, customers reported that they could overhear others "well". In RI , overhearing was between "some" and "well". However, in all of the eateries, on average, shown with squares, customers didn't report overhearing more than "some"; even in seniors' homes it was reported to be less than "some". Customers were asked to comment on any important issues that may have been missed by the questionnaire, or that they wanted to discuss. Out of 24 respondents in B l , six made comments. One complained about "too much echo...". One commented that they would prefer that radio was played instead of music. There was a complaint about noise after recent renovations: "this area was recently renovated and went from quiet to very noisy. There is a lot of reverberation and all conversation is amplified across the room." One customer reported her preference for seating location because of noise: "I would rather sit close to windows and far from the kitchen area." Finally, a different opinion about the noise was expressed by another customer: "It is nice to have a variety of noise and activity around. That gives some life to a dull place!" In B2, four customers out of 22 commented. Two complained about loud equipment: according to one, "It was pretty loud - the coffee maker, the fridge and also the adjacent table 41 with six male customers"; the other commented, "the fridge is a bit loud". There was a complaint about noisy chairs: "chairs were quite loud". However the music level was good, based on one customer's response: "nice music level and ambient" (note: the measured level of music in B2 was 65 dBA, equivalent to a 'normal' male-voice level). In B3/in, 8 out of 17 completed questionnaires contained comments. A l l complained about loud music in the E E . One customer wrote: "Turn down the volume on the music. It is regularly very loud. Also play calmer music (jazz-ish) vs. rock and roll". Another said: "Mostly, turn down music - every other form of noise rises with loud music", and one customer simply mentioned: "Use a little lighter music". One was concerned about the effect of loud music on her digestion system: "The problem is, the music is blasted way too loud. The level of noise is disturbing when I eat. I was told I do not digest properly because of the noise". There were suggestions, such as, "Turn down the music!", "Only have speakers in certain areas.", "Noise tolerance varies by time of day.", "Normally I prefer no music vs having a beer, when I like more music", "Speakers in the corner drop the sound on the tables. More surround sound is better here", and "Music is too loud. More absorbent materials for flooring or hanging on the wall. Somehow the combination isn't relaxing or forces me to concentrate to block out the wall of sound to tune my ears out, to avoid hearing specific conversations. This is quite tiring" (note: the measured level of music in B3 was 76 dBA, higher than a raised male-voice level of 72 dBA). In B3/out, two out of 17 made comments. One wrote, "Music sucks", and the other, "I like noise unless I am studying or in a meeting". In R3, six of 19 completed questionnaires contained comments. One customer reported the importance of this survey, and added: "In restaurants background music should be loud enough to distinguish the relevant point of identification or it is noise. If music is too loud it severely detracts from enjoyment. A recent visit to the WaaZubee cafe led me to pay the wrong bill because I was so intent on escaping and the waiter brought the wrong check. The music was very loud. This survey is important" (note: the measured level of music in R3 was 58 dBA, similar to a casual male-voice level of 59 dBA). Two hard-of-hearing customers commented on noise: "Background noise in public places makes hearing virtually impossible, despite a hearing aid", and, "I wear a hearing aid and I am thus sensitive to these issues. My seat location at today's meal was exceptionally good (close to the south private area and window...) but, in fact, the only location that seriously impedes conversation is near the bar. Loud background music would be disastrous". One customer complained about group talking that caused additional 42 noise: "a large group at one table can cause noise". Two of the customers who participated in a dinner meeting complained about H V A C noise: "During the dinner meeting the H V A C system was really noisy", and the other, "Fan and music should be turned off for business meetings because of the poor acoustical conditions of the place". In C l and C2, five customers out of 44 made comments. One customer in C l and another in C2 complained about noise due to group talking: "Usually noise comes from groups of people that cluster their tables together", and, "Gets very loud when (crowds) more than five people dominate one area". Two of the respondents in two different areas requested music: "I think having music for a certain section in the eating area is great", and, "I like background music; light classical or jazz or even talk radio". However, the acoustical environment was considered satisfactory by one customer: "Loud noises normally bother me if it is a quiet room, but I expect this cafe to be loud and it is fine". Note that there was no background music in C l and C2. In R l , four out of 15 customers made comments. Three complained about noise in different ways: "I bring my parents here once per week and I have always found the noise level here very high. We come here because it is an old favourite for my parents. My father is somewhat deaf and my mother has Alzheimer's. I am the only one who is bothered by noise", "Dish collecting was intrusive, sound echoed (high ceiling)", and, "Noise from outside is not an issue. Noise from customers should be dampened". One person complained about music: "Music played was instrumental, so it changed often in pitch, etc. Music affected me because it was changing from high to low too often" (note: the measured level of music was 62 dBA, similar to the normal male-voice level of 65 dBA). In R2, four out of 10 customers made comments. They complained about the noisy fan in the kitchen, and about the acoustics of the EE: "Fan in kitchen is loud", and, "The acoustics are really bad, especially when it is full". However, two were satisfied with the acoustical environment: "The restaurant has a low 'buzz' and general sounds are not intrusive", and, "Since the installation of the carpet, the noise level has improved". In S2, just one resident out of six made comments. However she didn't complain about the acoustical conditions, simply noting that: "seniors are not noisy". In SI, five customers out of 16 made comments. They complained about noise when it is busy; however, they didn't find the place worse than other restaurants. Two of them said: "Noise is predominantly from other voices", and, "Acoustically, I consider the dining room average, but staff make it super. The 'din' of conversation is a negative factor, but it is caused by my hearing aids and is no worse than any other restaurants: If tables near me have people with loud voices, 43 it bothers me. If voices are 'normal', it causes very little problems". Hard-of-hearing residents found the acoustical environment not to be optimal: "Because of the increased numbers using the dining room, the background noise level can be high, often impairing ability to hear conversation at the table, especially with those who are hard-of-hearing", and, "Hearing aids are good in a quiet room; however they should be removed in a noisy place". One apparently sent a group request for more acoustical treatment: "Some agree that sound insulation is necessary to provide adequate sound control". 4.4 Employee questionnaire A total of 27 completed questionnaires from employees, 21 females and 6 males, were analyzed. The number of completed employee questionnaires varied from 1 in B l to 6 in R2, with an average of 3. The job titles of the employees who participated varied, and included: manager, supervisor, cashier, waiter/waitress, bartender and cook. Employee ages vary from 21 to 55, with an average of 35 years. Their working experience varied from 0.3 to 19 years. Workers on average worked for 27 hours per week in eateries, and 34 hours per week in seniors' homes. Twenty-six percent of the employees reported that English was not their first language. About 26% of the employees reported being aware of having a hearing impairment; 14% reported that this was "moderate", the rest "mild". Three of four employees in B2 reported experiencing a hearing problem since they were hired. 44 Employees were asked about bothersome sources of noise. As shown in Fig. 22, on average, kitchen equipment and activities were the most annoying sources of noise for employees. Customer's moving, colleagues talking and moving, kitchen activities and equipment bothered about 70% of the employees in the seniors' homes. Since there was no music in the seniors' homes, and cell phones were not allowed in the dining rooms, there were no complaints about those sources of noise. However, cell phones were one of the most annoying sources of noise for employees in eateries. Employees were asked to indicate to what extent sources of noise bothered them in the work- place. As shown in Fig. 23, employees in seniors' homes were bothered more than in eateries by sources of noise such as customers moving, colleagues talking and moving. In general, employees in seniors' homes were more conscious of others and the noise they generate compared with eateries. However, in eateries employees were bothered more by kitchen activities and equipment. 4 5 W®CrJ7 DDflKKgOo ©©ODD® Noise sources Figure 23: Extent of bother of employees by different sources of noise: ( — - ) eateries, ( ) SHs. Employees were asked about problems that they experienced due to the noise in their working area. As can be seen in Fig. 24, at least 70% of the employees in eateries had difficulty talking with their colleagues, and needed to raise their voice to be able to communicate. However, in seniors' homes, about 60% of employees experienced "health-related" problems such as tinnitus, fatigue and headache, as well as broken concentration due to the noise. Because of the noise in seniors' homes, almost 50%> of workers had difficulty talking with their colleagues and residents. As seen in Fig. 25, all these problems were experienced not more than "some" in EEs. 46 Consequences Figure 24: Percentage of employees who experienced different problems due to the noise: ( • ) eateries, ( • ) SHs. MafiaQD Problems Figure 25: Extent of experiencing different problems due to the noise: ( ( ) SHs. ) eateries, 47 Employees were asked to indicate to what extent they could easily hear conversations at the tables, and between their colleagues, when they were not part of those conversations. About 83% of the workers in EEs reported that they could overhear conversation at the tables. However, 89% of employees in eateries reported that they could overhear their colleagues' conversations; the proportion in seniors' homes was 83% - however, not more than "some". Employees were asked about the acoustical environment of their working area and how well their expectations were met for an ideal work environment. On average, employees in EEs found acoustical environment to be "average". However, employees who were working in the kitchen area in B2 and R2 found their workplace to be "very poor" and "poor", respectively. Finally, employees were asked to comment on any important issues that may have been missed by the questionnaire. Cooks and prep cooks complained about noisy ventilation that contributed the most to difficulty in conversation, and that masked sound coming from the main eating area. In B3, in spite of loud music, and because of the noisy ventilation, the cook commented that she wished the music was played louder! 4.5 Customer questionnaire correlations To investigate possible relationships between customers' responses to different questions, and between physical measurement data and questionnaire responses, as well as to answer research questions established for this analysis, statistical correlation analysis was used. It was applied to each EE individually, and for all of them together; only the combined data is discussed here. A second aim of this analysis was to establish groupings of questions that could be considered to combine into a single factor or variable representing part of the overall subjective interpretation of the acoustic environment. Values of the correlation coefficient from 0.5 to 0.999 were considered to represent a strong direct correlation. Values between -0.5 and 0.5 were considered weak correlations. Values from -0.5 to -0.999 were considered to represent strong inverse correlations. Following is the interpretation of the correlation results for each research question - the objective here is to highlight interesting apparent relationships, not necessarily to identify causation. 48 > How are the reasons for visiting EEs related? The correlation results suggest that customers who frequent eateries tend to go there for two different combinations of reasons. Some visit for business, studying/working, celebration and relaxing, and others for eating/drinking, talking and relaxing. Residents in seniors' homes frequent EEs mostly for eating and drinking. The more residents in seniors' homes visited the EEs for business and celebration, the more they found the place to be good for talking. t > Are there relationships between the EE activities and the consequences customers experience? The correlation results suggest that customer enjoyment of an EE is reduced when their expectations are not met for eating/drinking and talking, when they can't hear others at their table and when they are bothered by music. Customers who are bothered by noise sources such as people speaking and moving, adjacent-table conversation, cell phones, kitchen equipment and activity, clinking of dishes, outside noise and music, suffer broken concentration, tired voices, headaches and fatigue. However, seniors experienced sources of noise such as cell phones, kitchen activities and equipment, outside noise and music less and, consequently, experienced less "health-related" problems such as headaches. > Are there relations between the noise sources in EEs? In EEs, people moving and talking, adjacent-table conversation, and noise from kitchen activities and equipment were experienced together as the most annoying sources of noise. However, in seniors' homes, reverberation related to people speaking is also a bothersome source of noise. > Are there relations between the various consequences of noise? ' In all EEs, difficulty hearing at the table due to noise was the predominant problem, and that was associated with customers experiencing broken concentration and tired voices. Customers experienced fatigue related to tired voices and headaches and, consequently, their enjoyment of eating out was reduced. In seniors' homes, the correlation between these problems was stronger. People speaking and adjacent-table conversation caused seniors to have difficulty 49 hearing at their table, difficulty talking with the waiter or waitress, and to experience broken concentration and tired voices. > What affects speech intelligibility and speech privacy? The results suggest that the more customer expectations are met for talking, the less they have difficulty hearing conversation at the table, and the more they enjoy eating out. Customers who report being overheard report overhearing, and vice versa. In eateries, the greater is a customer's hearing loss, the less they overhear other tables, or are overheard by them; however this wasn't found in seniors' homes. In eateries, it has been found that when speech intelligibility is impaired, customers have difficulty talking and hearing at their own table and to waiters, and that people speaking, outside noise and music impair speech intelligibility. However, in seniors' homes people speaking is the main problem. > How do the EEs' physical and acoustical characteristics affect customers? The correlations between the physical measurement data and the questionnaire responses, suggest that the higher is the noise level in the occupied EE, the less seniors visit. The higher is the noise level in the occupied EE and the greater is the seat density, the less residents in seniors' homes were bothered by kitchen activities and clinking sounds. With increasing average occupancy, adjacent-table conversation bothered residents in seniors' homes more. In all eateries and seniors' homes, the more customers reported difficulty talking at their own table, the more they had difficulty talking to waiters, and the more they suffered broken concentration and tired voices. > What is related to hearing loss in EEs? The results suggest that residents in seniors' homes who wore hearing aids didn't find those EEs to be good for eating/drinking, talking and celebration. However, outside of seniors' homes, in eateries, customers with more severe hearing loss are more likely to visit the place for business and celebration, and report fewer customers in the place compared to normal-hearing customers. They visit EEs which have more sound absorption, lower noise levels and which are larger. They prefer quiet and no music. They are bothered more by reverberation, and are less 50 bothered by kitchen-equipment noise. The more severe is their hearing impairment, the less they feel they overhear others, or are overheard by others. 4.6 Employee questionnaire correlations Employees were asked about the acoustical environment in their workplace, such as the sources of noise and problems they experienced due to the noise, as well as personal questions such as age, sex, hearing ability and their typical working area. At the end of the questionnaire, they were asked to comment on any acoustical issues that the questionnaires missed. Correlation analysis of the data was done; the results interpreted in the light of relevant research questions are presented here. Following is the discussion for each research question. > Are (here relations between how the sources of noise bother employees? Noise sources that bother employees include customers' and colleagues' talking and moving, kitchen activity and equipment associated with H V A C noise. The results suggest that workers, such as waiters, who work in the main eating area, are more likely to be bothered by customers' moving and talking; however, those employees who work in the kitchen area found kitchen equipment, activity and H V A C noise to be most annoying. > Are there relations between the different noise sources in EEs? Bother from customers moving and talking and colleagues moving and talking is related to bother by outside noise. The reporting of kitchen activity and equipment as a source of noise, and the extent of bother from these sources of noise, are related. Because of the noise from kitchen equipment, cell-phones (ringing or talking) are not a problem. H V A C noise as a component of kitchen-equipment noise is a problem. > Are there connections between the EE activity and the consequences employees' experience? Employee reports of frustration were associated with experiencing colleagues talking as a source of noise. The consequences of outside noise were headaches for employees; that of noise from kitchen equipment was tinnitus. Employees experienced headaches, fatigue and tinnitus due 51 to customer movement. The more colleagues talking was reported, the more employees suffered frustration. The older the employee, the less cell phones bothered them. > What is related to hearing loss for employees in EEs? The more severe is an employee's hearing loss, the less they are bothered by customers' talking; however, the longer they have had their hearing loss, the more customer moving and kitchen activities bothered them. The longer they have suffered from hearing loss, the more they reported the need to raise their voices in conversation. Employees who had suffered hearing loss for a longer time, reported overhearing customers' conversations less. > How do the EE physical and acoustical characteristics affect employees? The higher is the level of background noise, the more cell phones are reported as a source of noise. The less is the background noise, the more customer and colleague movement is noticeable and bothersome for employees. There is a strong direct correlation between the level of H V A C noise and the employee noise exposures. The greater is the background-noise level in the occupied EE, the more employees need to raise their voices (Lombard effect). The bigger are the area and volume of the EE, the more employees report difficulty talking with customers. Difficulty talking to customers and colleagues because of noise are related. The higher is the reverberation time, and the bigger are the room volume, surface area and volume-to-surface-area ratio, the more difficult it is to talk with customers; the bigger is volume, the more difficult is talking to colleagues. > How do the EE acoustical characteristics affect fob performance of employees? The more customers' talking bothers employees, the more they have problems talking to customers, and the more they suffered from broken concentration and frustration. The more they are bothered by outside noise, the lower they rate the quality of their workplace. > How does the EE acoustical environment affect employee health? The more employees were bothered by customers moving, the more they experienced fatigue and headache. The more they were bothered by colleagues talking, the more they experienced broken concentration and frustration. The more they were bothered by noise from 52 kitchen activities, the more they had problems talking to colleagues, and suffered broken concentration and frustration. The more they were bothered by kitchen equipment, the more they experienced broken concentration, frustration and tinnitus. The more they were bothered by outside noise, the more they experienced headaches. Employees who reported difficulty in talking with customers and colleagues reported experiencing headaches, fatigue and tinnitus. Those employees who needed to raise their voices to overcome the noise, experienced frustration. The greater is the number of hours of working per week, the more employees experienced fatigue and headaches, and the more they have difficulty talking with colleagues, related to the noise. The more employees have difficulty talking to their colleagues, the more they need to raise their voices, and the more they feel frustrated. 4.7 Discussion Customers visit EEs for more reasons than just eating/drinking and talking. Many visit for working/studying, business, celebration and relaxing. People visiting EEs tend to fall into two groups, one which visits for eating/drinking, talking and relaxing, the other for working/studying, business, celebration and relaxing. Generally EE customers don't enjoy their visit if there is inadequate speech intelligibility between people at their table, and inadequate speech privacy between tables. Lack of speech intelligibility at the table causes customers to suffer broken concentration and tired voices. The likely reason for the tired voices is the need to raise the voice to be heard over the noise (Lombard effect). Customers who reported overhearing, reported being overheard more. However, people with more hearing loss reported less overhearing and being overheard. Customers prefer to dine in a quiet environment with an appropriately low level of music. In EEs with loud sources of noise (kitchen equipment, H V A C , loudspeakers playing music), customers choose their seats to avoid them. Customers suffer various problems due to noise, including reduced enjoyment, difficulty talking and hearing, broken concentration, tired voices, headaches and fatigue. However, they experience these problems not more than "some". The most annoying sources of noise are other people talking, kitchen activities and equipment, and cell phones. Although these sources of noise are bothersome, and impair conversation, they are usually not experienced more than "some". Excessive reverberation is also a problem. 53 The expectations of hard-of-hearing seniors who visit EEs for socializing and relaxing are generally better met than for the normal-hearing. This can be explained as follows: they don't experience sources of noise such as cell phones, kitchen activities and equipment, outside noise and music as much as the normal-hearing customers, and their expectations of EEs are to be somewhat noisy. On the other hand, since they prefer to dine in a quieter place with lower background-noise level, smaller volume and surface area, and with lower average occupancy, they perceived less noise. For employees, people talking and moving, and kitchen activity and equipment, are the most annoying sources of noise. These sources of noise cause them to suffer from fatigue, tinnitus and headaches. Cell phones are more bothersome with increasing background-noise level. However, cell phones (ringing or talking) are not a problem for older and hard-of-hearing employees. Significant relationships are found between working hours, being more concerned about perceived noise levels in the workplace, and experiencing health problems. The greater is the number of working hours per week, the more employees experienced fatigue, headaches and difficulty talking with colleagues and, consequently, frustration. Hard-of-hearing employees report less difficulty talking with colleagues and customers. However, the longer they have been hard-of-hearing and the more severe it is, the more they need to raise their voices to be heard, and the more they suffer broken concentration. The severity of hearing loss can result from a longer period of exposure to the noise. The greater is the background-noise level in the occupied situation, the more employees need to raise their voices to converse. 54 Chapter 5 ADVANCED NOISE ANALYSIS In Chapters 3 and 4, measurement data and questionnaire results were discussed. In this chapter, measured noise levels are investigated further, with the following objectives: > to identify the relationship between noise level and occupancy (the number of customers) at a given time in EEs; > to investigate the typical voice levels of talkers in EEs; > to evaluate typical acoustical conditions for speech intelligibility in EEs by way of signal-to-noise ratio calculation and existing acceptability criteria; > to investigate how much customers need to raise their voices with increasing noise level in EEs (the Lombard Effect). In order to identify the relationship between noise level and the number of customers at a given time, the customer questionnaire responses and noise measurements were subjected to further analysis. The customer questionnaires asked about their arrival and departure times in the EEs, and the number of people in the place while they were having their meal. The number of customers was calculated as the average response of customers to this question during the same period of dining (arrival to departure). As discussed in Chapter 3, the equivalent sound pressure level ( £ e q ) 6 0 s ) had been recorded every 60 s during operation hours using noise dosimeters. Equivalent sound pressure levels for each dining period (Lcqpcr) were calculated as, where n is the number of measured Icq^Os values during the dining period, and T is the duration of the dining period in s. Equivalent sound pressure level was plotted against the number of customers for each EE, as shown in Fig. 26. (?) I=I 55 C2 # of customers gure 26: Variation of LeqpeT (dBA) with number of customers in ten EEs, and the logarithmic gression lines through the data points. 56 The pooled data for all EEs is shown in Fig. 27. Also shown in each case are the logarithmic regression lines through the data. As expected, noise levels increased with increasing number of customers in all EEs, but with the different slopes, with the exception of R3 and C2. The decrease of noise level with increasing number of customers in C2 could be due to the nature of the place and its customers. A l l of the respondents in that place visited the place with the objective of relaxing. The objective of relaxing could result in customers not communicating with others as much as in the other EEs, and mostly concentrating on individual work. The different slope in R3 could be due to the 'granularity' (non-uniformity) of where customers sit, since this EE was bigger than the others. Other reasons could be the location of the noise dosimeters, and that not all of the customers were talking at the same time. 8 0 75 < 7 0 CD I J 6 5 6 0 55 • • • • • • 10 # of customers 100 Figure 27: Variation of pooled Lcq,pcr with the number of customers with all EE data pooled, and the logarithmic regression line. 57 In the pooled data in Fig. 27, Z eq , Per increased with an increasing number of customers. However, due to the different acoustical environments, different occupancies and different talker voice levels in the different EEs, a large scattering of the data can be observed.In order to investigate the typical voice levels of customers in EEs, the noise levels due to the occupants were calculated. B N L was energetically subtracted from Zeq,per to give Zeq,corr- The measured B N L , the noise level in the unoccupied EE, comprises noise from equipment, external noise sources and the H V A C system, measured with a sound level meter, as discussed in Chapter 3. Next, using Eqs. (1) and (3), the total sound-power level of the customers Z,W;tot was calculated from Z.eq,corr under the assumption that the noise dosimeter was far enough from talkers so that only the reverberant field need be considered, as follows: Iw,tot=^,corr-101og(4-),dBA (8) R in which R = aS/(\-a) is the room constant, with a and S the EE average surface-absorption coefficient and total surface area, respectively, as discussed in Chapter 1. Under the assumption that all of the occupant noise is from talkers' voices, and that the number of talkers is one-third of the number of customers (this will be discussed further below), the sound-power levels or voice levels of each talker Z„,t in each dining period in each EE were calculated, as follows: Lwx = Z,,,tot - 10-log (# of customers / 3), dBA (9) The variation of Lwt with the number of customers is plotted for each EE in Fig. 28. Also shown are the power levels associated with different reference voice levels (Normal, Raised, Loud and Shouting), taken from ANSI S3.5, and shown in Table 1. Casual levels have also been estimated and are included. Casual to Normal voice level was expected to use by customers in quiet. On average, customers used Casual to Raised voice levels for conversation in EEs, with some exceptions. Due to the low background-noise level in the S2 seniors' home, a Casual voice level was used for conversation. On the other hand, on average, a Raised to Loud voice level was used in Bistro B3 due to the high background-noise level (loud music). 58 Table 3. A N S I S3.5 reference mean octave-band free-field sound-pressure levels at 1 m on-axis and sound-power levels for speech at five vocal efforts. Frequency (Hz) 250 500 l k 2k 4k 8k d B A LPm (dB) Casual 52.0 53 47.5 42.8 37.8 37.0 54.0 Normal 57.2 59.8 53.5 48.8 43.8 38.6 59.5 Raised 61.5 65.8 62.3 56.8 51.3 42.6 66.5 Loud 64.0 70.3 70.6 65.9 59.9 48.9 73.7 Shouted 65.0 74.7 79.8 75.8 68.9 58.2 82.3 Casual 60.0 61.0 55.5 50.8 45.8 45.0 63.0 Normal 67.3 68.7 61.5 55.8 50.0 43.6 68.0 Raised 71.5 74.6 70.3 63.8 57.5 47.5 74.8 Loud 74.1 79.3 78.6 72.9 66.1 53.8 81.7 Shouted 75.0 83.7 87.8 82.8 75.1 63.2 90.0 To evaluate the acoustical conditions for speech intelligibility in the EEs , speech levels Z s , equal to the free-field sound-pressure levels at l m , Lpm, were calculated from Z w t , as follows: U = L p in Z w t - 20 log (r) + 10 log ( 0 - 1 1 = Z w . - 8, d B A (10) In this equation r, the distance between the talker and the listener, is assumed to be 1 m (a typical table width) and the directivity of the talker Q is assumed to be equal to 2, as discussed in Chapter 1. Considering the above inputs for Q and r, Ls values are 8 dB less than the corresponding L w t values. A-weighted signal-to-noise ratios, SNA, at the listener positions were then calculated as the difference between and the speech-signal level Ls and the noise level Ln, calculated by energetic subtraction of the speech level from the total sound level Z e q , p e r ; L„ =10 log 1 0 A ( ^ ^ ) - 1 0 A ( ^ ) 10 10 d B A 01) 59 90 80 70 60 -F B1 R3 90 80 70 60 -!-10 R2 HP*-10 JL JJ3 100 100 JO 100 90 80 70 60 90 80 70 60 90 80 70 n. -m © T ^ H IL IL 60 4 10 C1 10 C2 10 100 100 100 # of customers Figure 28: Variation of talker voice power level Lwt with number of customers in individual EEs. Also shown are logarithmic regression lines and reference voice levels. (C = casual, N = normal, R = raised, L = loud). \ 60 55 60 65 70 75 80 55 60 65 70 75 80 55 B3 60 65 70 75 80 12 8 4 0 -4 -8 -12 55 C2 60 65 70 75 80 55 R2 • 60 65 70 75 80 12 8 4 0 -4 -8 -12 55 S2 60 65 70 75 80 Leq, per (dBA) Figure 29: Variation of SNA with Lt%va in individual EEs, with linear regression lines. SNA values were plotted against Z e q tot, as shown in Fig. 29. Falues varied from -12 to +10 d B A and were, on average, negative in all EEs, with some exceptions ( C l , SI , S2). Average values varied between -2.7 in R3 and +2.1 in S2, as shown in Fig.30. This range of signal-to-noise ratios is lower than the minimum ratio of about 5 or 6 dB that is required for face-to-face talking when facial expression and gestures contribute to intelligibility [19]. To investigate the Lombard effect in the EEs , Lwt was plotted against Z eq, Per, as shown in Fig. 31 for the pooled data. Clearly, as noise levels increased, customers needed to raise their voices to overcome noise in the EEs. The corresponding Lombard rate, the increase in voice level LwX for 1 dB increase of noise level ^eq.per was 0.7 dB, as shown from the slope of the regression line in Fig. 31. The Lombard rate was reported to be 0.2 to 1 dB/dB in the literature [28-35]. A model for predicting speech and noise levels, and the acoustical conditions for speech intelligibility, in EEs including the Lombard Effect w i l l be discussed next. 20 -15 -10 -< -5 • • • • • • • • * ? — • • • •• • • • • — — 2 • — • ~ o * • • • 1^* • • — • 55 60 65 70 75 80 Leq , per(dBA) Figure 30: Variation of S N A ( d B A ) with Leq, per(dBA) in all EEs, and the linear regression line. 62 < 50 55 60 65 70 75 Leq .pe r (dBA) 80 Figure 31: Variation of talker vocal power level Z„, t with Zeq,Per in all EEs, and the linear regression line and equation. 63 Chapter 6 PREDICTION In order to find the best way to optimize the acoustical environment in EEs, the C A T T -Acoustic software was used to predict Speech Intelligibility (SI) and Speech Privacy (SP) at a typical table. SP is essentially the opposite of SI. When the background noise is high relative to the speech level and/or the reverberation is excessive, then the speech wi l l not be intelligible; thus SI is low, but SP is high. To evaluate SI and SP at a typical table in EEs , STI and its components, EDT and SNA were predicted with CATT-Acoust ic . The EDT and STI were discussed in the Chapter 1. SNA is the A-weighted speech-to-noise level difference. It is calculated from the predicted speech (Z s) and noise (Z n) levels. In general SI tends to increase with increasing SNA. To include the Lombard effect and its variation with E E occupancy in all predictions, talkers' voice levels predicted by the "Lombard prediction" model developed by Hodgson, Steininger and Razavi [37], were used. The "Lombard prediction" was developed based on diffuse field theory, with a correction factor that corrects the reverberant sound-pressure level for non-diffuseness. A s discussed before, diffuse field theory can only be accurate in an empty, quasi-cubic shape room that has diffusely reflecting surfaces with uniformly distributed absorption. CATT-Acoust ic was used to take into account the non-diffuseness of the sound field in an E E due to the non-cubic, non-uniform absorption and diffusion distributions and the presence of barriers. 6.1 CATT-Acoustic CATT-Acoust ic is a room-acoustic prediction model based on ray-tracing and the Image Source Model (ISM), as discussed in the Chapter 1. This model requires three different files defining the input data: the geometry of the room, including the absorption and diffusion coefficients of the different room surfaces; the locations and powers of the noise source(s); and the number and locations of the receivers. Creating an A u t o C A D file of the E E and importing it into CATT-Acoust ic with information about absorption and diffusion coefficients and BNL defined in octave bands, impulse responses are predicted. From the impulse responses relevant acoustic parameters (e.g. RT, EDT, Lp) characterizing the acoustical environment are predicted. 64 Lp is the sound pressure level of a source at the receiver position. STI values were calculated with CATT-Acoustic based on the Houtgast and Steeneken [17] method, as discussed in Chapter 1. 6.2 "Lombard prediction" model t To predict talkers' voice levels at different occupancies, including the Lombard Effect, an iterative model proposed by Hodgson, Steininger and Razavi [38] was used. This model is based on using measured noise levels and questionnaire responses, with some assumptions about the number of talkers per customer, and diffuse field theory. However, in this prediction a correction factor for non-diffuse sound field was considered, to correct for different geometries and non-uniform absorption distribution, but not for barriers. A hypothetical EE with dimensions L, W and H, surface area S, average surface-absorption coefficient a, room constant R and a uniform BNL was considered. It was assumed that customers visited the place in pairs, talker and listener. The listener is in the direct field of the talker (the contribution of the reverberant field is negligible). Different pairs are in each other's reverberant field (the contribution of direct field is negligible). The total number of customers at the EE is NC = N, * TPC, where N, is the number of pairs of talkers and listeners and TPC is the number of talkers per customer. In the absence of BNL, each talker would talk with a free-field sound pressure level at 1 m, equal to L Pfn, q. Letter q stands for quiet. However, due to the inanimate noise (BNL) in the EE, based on the Lombard Effect, they need to raise their voice and talk in a voice level at noise, Lpiri,,!, louder than Lpffi ,q. To predict the voice level in noise when the Lombard Effect occurs, the following Lombard model was proposed, with some unknown Lombard parameters: asym 1 + exp xmid - Ln scale dBA (13) The behaviour of this model and parameters asym, xmid, scale and also L p fn ; q and Lpffi j m a x are illustrated in Fig. 32. The reason for proposing this equation was to find an S-shape curve that fits the measured data the best. S-shape curves could have different shapes and the logistic curve 65 is the most flexible one, and so it was used in this model. An S-shape variation of voice level with noise level is appropriate, since people start talking in a voice level of LPfn,q and increase their voices with increasing noise until the voice level converges due to the vocal chord limitations. The model assumes that voice output levels vary between a minimum of L p fn j q and a maximum of Lpffi ; m a x = Lpffi,q + asym, and that the voice output levels vary at a rate of asym/(4*scale) dB/dB between these two limits. Thus asym is the difference between the maximum voice level and voice level in quiet. Scale determines how steep the slope is. Xmid, as shown in Fig. 32, is the voice level between the two different of lowest and highest conditions. The corresponding voice power level at noise, knowing the directivity of human voice, will be Av,t = Lpm,n + 8. Under the assumptions made above, the speech level at the listener is Ls = Lpm,n-The reverberant pressure level that each talker generate equals Z r e v , t = Lw>t + 101og(4//?) in which R is the room constant as explained in the Chapter 5. Since this expression is derived from diffused field theory, a correction factor of RCF, is added to compensate for the non-diffuseness of the sound field. The total noise level recorded with a noise dosimeter, L Q , in the reverberant field of all talkers is as follow: Note that © indicates a decibel sum. However, each talker experience a noise level, L„, due to the reverberant level of the others, as follow: This is also the noise level Ln at each listener, so they experience a total, A-weighted signal-noise level difference SNA = Z s - Ln. This prediction model can be used to model speech and noise levels in an EE containing any number of customers/talkers, including the Lombard Effect. However, it is based on diffuse-field theory. Moreover, it contains a number of parameters for which the values are not known for EEs. LD = B N L © [Z r e v > t + 101og(M)], dBA (14) Z n = B N L © [Z r e V lt+ lOlog(M-l)], dBA (15) 66 xmid Ln (dB) Figure 32: Proposed Lombard model used in this thesis. These include the parameters defining the Lombard Effect - LpmA, asym, xmid, and scale in Eq. (13) — and parameters defining the EE, some of which can be assumed to be the same in all EEs (Ay,, TPC) and some which would be expected to vary from EE to EE (L, W, H, a, BNL, RCF). Av is the absorption coefficient per person, as discussed in the Chapter 1. For each of the EEs studied in this thesis, data derived from measurements in them (the average dining-period noise level, Zeq.per, and the sound power per talker, Lw{) were input as the target outputs of the prediction model. Optimization techniques were used to estimate the values of the twelve unknown input parameters for each of the EEs. Note that values of several EE parameters - in particular, L, W, H, and a - are nominally known, since they were measured. Thus, the option exists to use the nominally-known values in the optimization, or to allow them to vary and be found. The expectation is that the optimization procedure would find values close to the measured values. Four different methods were used to find optimal values for these twelve parameters: 'fixed, constrained', 'fixed, unconstrained', 'constrained' and 'unconstrained'. Several different techniques were used to choose the starting values for the optimization procedure. L, W, H, a, and BNL were chosen to be the measured values. Z^rrui , RCF, TPC and Ap were chosen heuristically. aysm and xmid were estimated from preliminary work with the Sutherland, Lubman and Pearson Lombard-Effect model [34]. The RCF varied from -5 to 5 to find the best value that fits the S shape curve and measured data the best. Finally, scale was chosen to take a value larger than it was reasonable to assume that it could be found to be. This was because it was observed that the interactive algorithms had more difficulty increasing scale 67 than decreasing it. Since there were different errors associated with the different optimization method, the values that were used in this thesis were average values taken from the 'fixed' cases which had relatively low errors. The optimal values defining the Lombard model, along with the optimal values for RCF, TPC and Ap, are tabulated in Table 4. The predicted talkers' voice levels from this model were used as inputs into the CATT-Acoustic software. 6.3 M e t h o d An EE of simplified geometry, with the same volume, surface area and average absorption coefficient same as one of the studied EEs, was considered, as shown in Fig. 33. The materials used in the predictions for the walls, ceiling and floor were based on the materials in the EEs, with some changes to obtain the same RTs as in the measurements. An omni-directional sound source was assumed for the simulation of reverberation times. The source was chosen to have an output free-field sound-pressure level of 90 dB at 1 m in octave-band frequencies from 125 Hz to 8kHz. STI and its component EOT and SNA values, for a typical seating position at different receiver locations in the hypothetical EEs, were then calculated. Table 4. The optimal values defining the Lombard model for the EE studied. Lpm^ asym xmid scale RCF T P C 56 25.5 70.5 • 1 -1.3 2.8 0.5 68 Figure 33: Simplified geometric shape of BI showing the layout of the tables; spheres are receiver position at a typical seating position; the square box is the source (talker). EDT values were predicted; since they are more related to speech intelligibility than RTs. Different numbers of customers were considered, based on for Low Occupancy-LO (25% occupied) and High-Occupancy-HO (75% occupied) in EEs. Correspondingly, the area per 2 2 person was 7.5 m for the former and 2.5 m for the latter. The suggested area per diner in the architectural design guidelines is approximately 1 m [39]. The number of the talkers was based on the "Lombard prediction" assumption of one third of the customers talking at the same time. For each level of occupancy, average absorption coefficients were calculated and put into the "Lombard model". In order to use the "Lombard prediction", the known values of quantities measured in the EE such as L, W, H, a and BNL were entered in the model. The BNL was equal to 60 dBA, as measured. The noise level at the listeners was equal to the sum of the B N L due to the inanimate sources of noise (equipment, fridge, outside noise,...) and noise due to the secondary talkers (customers seated at other tables from the listener). The talkers' voice levels were based on calculated voice levels, L p in , n , that each talker used during dining, with the assumption that the voice level was Lpmq in quiet, as explained in the Chapter 5. A l l talker and receiver positions were based on actual seat locations in the studied EE, and were at the typical height of 1 m. The distance between the primary talker (seated at the receiver table) and the receiver was considered to be lm, same as the table width. The distribution of talkers voice level with octave-band frequency was considered to be similar to that of the human speech spectrum. 6 9 The talkers' voice directivity was based on default values defined by CATT for human talkers, at the 125-4000 Hz octave-band frequencies, it was: 0.1, 1.1, 2.1, 3.1, 4.1, 5.1 dB. STI values were calculated with CATT using the noise level at the receiver position and the voice level of the primary talker which was the same as the other talkers. The STI values along with their corresponding SI are tabulated in Table 5. The average STI at the other receiver positions represents Speech Privacy (SP) at the typical table. Values of STI less than or equal to 0.2 provide acceptable SP. The effects of various absorptive materials on different surfaces, such as absorptive ceilings, walls and floors, were determined. Since the ceiling is the most important reflecting surface in rooms, it is important that it be highly absorptive. In open-plan offices, it is suggested that using thin carpet or thick carpet on the floor doesn't make a noticeable difference in speech intelligibility [40]. However, a non-absorbing floor does decrease the speech privacy. Predictions with thin carpet were made. The effect of barriers between tables was also predicted. The main reason for inserting barriers was to increase the early energy by adding more early reflections and to attenuate noise from other tables. To avoid increasing the late energy, the ceiling was made absorptive. Barrier heights considered were high enough to block the direct sound from one table to the others, and also sound diffracted over the barrier. Since talkers and receivers were considered to be at 1 m above the ground, the height of the barriers should be greater than this. Predictions were made with three different barrier heights of 1.2, 1.5 and 1.7m. Two different barriers absorptions were considered, non-absorptive and high absorption with a equal to 0.08 and 0.75, respectively. The following abbreviations were used in this chapter: UC: Untreated Configuration U C B : Untreated Configuration with Barriers A C : Absorptive Ceilings A W : Absorptive Walls AF: Absorptive Floor A C B : Absorptive Ceilings with Barriers A B : Absorptive Barriers L A B : Low Absorptive Barriers N : Normal voice level 70 R: Raised voice level L: Loud voice level S: Shouted voice level. 6.4 E E configuration EE BI was chosen as a model with dimensions of 16 m length, 10 m width and 4.5 m height. Materials for the floor, wall and ceiling were selected from the CATT database based on the actual EE materials in B I : painted plaster surfaces for the walls, wooden floor for the floor and ceilings, and double-glazing (2-3 mm glass with 30 mm gaps) for the windows. Tables were added to the EE to add diffusion and to obtain predicted RT values equal to measurement. The diffusion coefficients of the room surface in six octave band frequencies from 125 Hz to 4kHz were the default values defined by CATT: 0.01, 0.03, 0.06, 0.1, 0.2, 0.6. Fig. 34 shows the receiver positions, numbered from 0 to 15, and the talker position, A6. Predictions were made in different configurations. It started with the Untreated Configuration (UC) and then the materials for the ceilings (AC), walls (AW) and floors (AF) were changed one by one to add absorption. In the end, barriers were inserted with different height and a values. The barriers were inserted in untreated configuration (UCB), and then to A C (ACB). Table 5. The STI values along with their corresponding SI. STI <0.30 0.30-0.45 0.45-0.60 0.60-0.75 >0.75 SI Bad Poor Fair Good Excellent 71 ,zm, Y rjo [34rj8C3;: QiALi&5rj9a:! X Figure 34: Floor plan of computer model of E E B l showing the receiver positions (0-15) and talker position, A6. The dimensions are also given. The empty area is the customer traffic area without any seats. The material used for one wall was considered to be windows, as in E E B l and wasn't changed in all predictions. Table 6 summarizes all the material used, with brief descriptions, absorption coefficients, a and average coefficients of the place a after adding absorption. The U C was based on the actual materials used in E E B l . a for the absorbent walls and ceilings were considered to be 0.75 and 0.70, respectively, with corresponding a equal to 0.37 and 0.31. Predictions were made with thin carpet for the floor with a equal to 0.17; this increased the average absorption of the E E to 0.17. The same predictions were made after inserting barriers, as shown in Fig. 35. 72 Figure 35: Barriers in E E B I with the same distance from all the tables. Table 6. Materials used in the different prediction configurations, with the corresponding a and Name Material a a A W Perforated metal (13% open, over 50 mm fiberglass) 0.75 0.37 A C 15 mm ceiling tiles 0.70 0.31 A F Thin carpet 0.17 0.17 U C Walls: painted dry wall 0.16 0.16 Ceiling: wooden ceiling 0.12 Floor: wooden floor on joists 0.09 Windows: double glazing, 3mm glass, 10mm gap, framed window 0.05 Tables: plastic cover of tables, empty plastic or metal chairs, per m2 0.12 L B Barriers: plywood paneling, airspace, light bracing 0.08 A C B 0.30 U C B 0.40 A B Perforated metal (13% open, over 50 mm fiberglass) 0.75 A C B 0.30 U C B 0.27 s G l o b a l r e v e r b e r a t i o n t i m e SabT —S EyrT e EyrTg -O T-15 - - - A - - ' T - 3 0 T r e f Oct 125 2 5 0 5 0 0 l k 2 k 4 k Figure 36: Predicted octave-band RTs in the model EE when untreated. Tref is the measured RT. The other curves are the RTs predicted by diffuse field theory. 6.5 Resu l ts Measured and predicted octave-band RTs in untreated configuration are shown in Fig. 36. As can be seen, there was a good agreement between measured and predicted values. Thus the computer model is a good representation of the actual EE. Now lets discuss about STI prediction. With LO, as can be seen from Table 6, talkers' voice levels varied from 61.2 dBA to 62.4 dBA (normal to raised voice levels, see Table 3), which was higher than voice level in quiet, LPfn,q, 56 dBA. Noise level due to the other talkers and inanimate noise at the typical receiver position varied from 60.4 to 62.1 dBA in A C B and UC, respectively, higher than BNL of 60 dBA. A comparison between values of Lpm,n,Ls,Ln,SNA, STI and EDT in the different configurations was made and is tabulated in Table 6. The A-weighted signal to noise ratio (SNA) was approximately 2.5 dB in all configurations, with the exception in A C B and UCB for which the corresponding values were 5.9 and 1.4 dB, respectively. The EDT varied from 0.29 s in A W to 1.38 s in A F . The low SNA values caused poor Speech Intelligibility (SI) in most configurations. The STI varied from 0.51 to 0.61 in AF and A C B , respectively, corresponding to fair and good SI. Inserting barriers with low absorption in A C increased STI from 0.55 in UC to 0.61 with the 74 corresponding SI of fair and good, respectively. The SP at the typical table varied from 0.06 to 0.19 in A C B and UC, respectively. The SP in UC at low occupancy was marginally acceptable. With HO, the talkers' voice levels were 70.6 dBA (raised-loud) in the UC, whereas adding absorption to the place allowed the talkers to decrease their voice levels to 62.6 dBA (normal-raised), on average. In A W voice levels were the lowest, 62.5 dBA, compared with the levels in the other configurations. The noise level at the receiver position due the other talkers and BNL was 73.3 dBA in UC and 76.7 dBA in UCB. The decrease in noise level due to the absorption varied from 1.6 to 10.1 dBA, in A F and A C , respectively. The L A B in UCB resulted in more noise compared to UC, and consequently higher voice levels of the talkers (Lombard effect). The SNA value in the UC was -0.1 dB, adding absorption increased the value, on average, up to 2.3 dB and in A C B even increased more to 2.6 dB. The STI value in UC was 0.41, corresponding to poor SI. With A W and A C , the STI increased to 0.52 and 0.55, respectively. Correspondingly, the SI improved significantly from poor to fair by adding absorption. The increase in STI in the A C B configuration was less than A C ; however it changed the SI from poor to fair, too. Adding absorption to the floor with low absorption didn't change SI; however it increased SP, and the corresponding value changed from 0.15 to 0.10. SP was preserved in all configurations due to the high noise level. The EDT varied from 0.3 s in A W to 1.37 s in AF. The increase in occupancy (or decreasing area per person) decreased SNA by 2.9 dB in UC. On the other hand, adding absorption by A W increased the corresponding value by 2.4 dB and by 2.7 dB in A C B . It was assumed that the distance between talker and listener was lm in all predictions, as stated. 75 T a b l e 7. Pred ic ted acoust ica l va lues in di f ferent conf igurat ions w i th the ta lker- l is tener d is tance equal to 1 m, a l o n g w i th the changes due to occupancy : L O (12 customers and 4 talker) and H O (36 customers and 12 ta lkers) ; N = no rma l , R = ra ised, L = loud vo ice levels. uc LO HO Changes L„f f |. n/dBA - voice range 62.4 (N-R) 70.6 (R-L) 4.6 L s / d B A 64.9 73.2 8.3 L n / d B A 62.1 73.3 11.2 SNA I dB 2.8 -0.1 -2.9 EDT Is 1.19 1.20 0.01 STI (at # 5 receiver) / SI 0.55 / Fair 0.41 / Poor -0.14 STI (at the other receivers) / SP 0.19 0.15 -0.04 A C LO HO Changes L„m , , /dBA - voice range 61.3 (N-R) 63.2 (N-R) 1.9 LJ dBA 63.7 65.5 1.8 L„/dBA 60.9 63.2 2.3 SNA 1 dB 2.8 2.3 -0.5 EDT Is 0.62 0.60 -0.02 STI (at # 5 receiver) / SI 0.56 / Fair 0.55/Fair -0.01 STI (at the other receivers) / SP 0.14 0.13 -0.01 A W LO HO Changes Lpfri.n/dBA - voice range 61.2 (N-R) 62.5 (N-R) 1.3 LJ dBA 63.5 64.9 1.4 LJ dBA 60.9 62.6 1.7 SNA 1 dB 2.6 2.3 -0.3 EDT Is 0.29 0.30 -0.01 STI (at #5 receiver)/SI 0.55 / Fair 0.52 / Fair -0.03 STI (at the other receivers) / SP 0.13 0.12 -0.01 A F LO HO Changes L„m , , /dBA - voice range 62.3 (N-R) 69.3 (R-L) 7 LJ dBA 64.1 71.2 7.1 L„/dBA 62.1 71.7 9.6 SNA I dB 2.0 -0.5 -2.5 EDT Is 1.38 1.37 -0.01 STI (at # 5 receiver) / SI 0.51 / Fair 0.40 / Poor -0.11 STI (at the other receivers) / SP 0.18 0.10 -0.08 A C B LO HO Changes Z,„i'n. n/dBA - voice range 61.5 (N-R) 63.5 (N-R) 2 L s / d B A 66.3 67.9 1.6 LJ dBA 60.4 65.3 4.9 SNA I dB 5.9 2.6 -3.3 EDT Is 0.50 0.53 0.03 STI (at # 5 receiver) / SI 0.61 / Good 0.54/ Fair -0.07 STI (at the other receivers) / SP 0.06 0.04 -0.02 U C B LO HO Changes L„m , , /dBA - voice range 62.4 (N-R) 71.8 (R-L) 9.4 LJ dBA 67.1 76.7 9.6 L„/dBA 65.7 76.1 10.4 SNA I dB 1.4 0.6 -0.8 EDT Is 0.95 0.96 0.01 STI (at # 5 receiver) / SI 0.52 / Fair 0.45 / Poor -0.07 STI (at the other receivers) / SP 0.07 0.05 -0.02 76 The same predictions were made on the assumption that the listener moves closer to the talker in order to improve intelligibility. Table 7. Summarizes predictions with talker-receiver distance equal to 0.5 m. It can be seen that, at low occupancy, the SNA values increased significantly from 2.8 dB to 8.2 dB in UC. Correspondingly, the SI increased from fair to good. In A C B configuration, SNA increased by up to 9.8 dB. The EDT varied from 0.29 s in A W to 1.19 s in UC. In this prediction the STI values varied from 0.70 to 0.75 at AF and A C B , with the corresponding SI of good to excellent, respectively. At high occupancy, the SNA values varied from 5 dB to 7.8 dB in AF and A C configuration, respectively. The increase in SNA compared to the corresponding value with talker-receiver at lm, on average, was 5.3 dB. At both LO and HO, the STI increased for about 0.2, on average, which is significant. To investigate the effect of barriers further with HO, which is the normal occupancy in EEs, more predictions were made with different barriers heights and absorption coefficient, a . Table 8, summarizes different predictions with 1.5 m high barriers. Two different a values, low and high barriers in configurations, A C and UC were considered. Inserting low-absorption barriers (LAB) in UC increased talker voice levels by 1.2 dB, due to the increased noise of 2.8 dB. The EDT decreased compared with UC, which can be due to the more diffraction over greater number of surfaces. Inserting absorptive barriers (AB) decreased noise levels by 5.9 dB, and consequently SI improved from poor to fair. Inserting L A B in A C had more effect, and caused a change in talkers' voice level from 70.6 (R-L) to 63.5 (N-R). The most effective case was A B with A C which changed SI form poor to good compared to UC. To investigate the effect of barrier height, predictions were made with 1.2 and 1.7 m barrier heights with L A B in A C . There wasn't any significant change in any of Zpfn.n, ^s, Ln or SNA and, consequently in SI and SP. Using 1.7 m height barriers are more costly than 1.5 m barriers. A l l the results were with the assumption of talkers and listeners height of 1 m. The results with 1.2 m barrier height will change if the talker and listener heights were larger. Then using 1.2 m barrier height won't be a good treatment. 77 Table 8. Predicted acoustical values in different configurations with the talker-listener distance equal to 0.5 m, along with the changes due to occupancy: L O (12 customers and 4 talker) and H O (36 customers and 12 talkers); N = normal, R = raised, L = loud voice levels. UC L O H O Changes Zpfri.n / d B A - voice range 62.4 (N-R) 70.6 (R-L) 8.2 Z s / d B A 70.3 78.5 8.2 Z n / d B A 62.1 73.3 11.2 SNA 1 dB 8.2 5.2 -3 E D T / s 1.19 1.20 0.01 STI (at # 5 receiver) 0.73 / Good 0.64 / Good -0.09 STI (at the other receivers) 0.19 0.15 -0.04 AC L O H O Changes Zptri.n / d B A - voice range 61.3 (N-R) 63.2. (N-R) 1.9 Z s / d B A 69.2 71.4 2.2 Z n / d B A 60.9 63.2 2.3 SNA 1 dB 8.3 8.2 -0.1 E D T / s 0.62 0.60 -0.02 STI (at # 5 receiver) 0.74 / Good 0.73 / Good -0.01 STI (at the other receivers) 0.14 0.13 -0.01 AW L O H O Changes Zpiri.ii / d B A - voice range 61.2 (N-R) 62.5 (N-R) 1.3 Z s / d B A 69.0 70.4 1.4 Z „ / d B A 60.9 62.6 1.7 SNA 1 dB 8.9 7.8 -1.1 E D T / s 0.29 0.30 0.01 STI (at # 5 receiver) 0.74 / Good 0.72 / Good -0.02 STI (at the other receivers) 0.13 0.12 -0.01 AF L O H O Changes Zpfiu, / d B A - voice range 62.3 (N-R) 69.3 (R-L) 7 Z s / d B A 69.7 76.7 7 Z„/dBA 62.1 71.7 9.6 SNA 1 dB 7.6 5.0 -2.6 E D T / s 1.38 1.37 -0.01 STI (at # 5 receiver) 0.70 / Good 0 .59 /Fa i r -0.11 STI (at the other receivers) 0.18 0.10 -0.08 ACB L O H O Changes Zpirij, / d B A - voice range 61.5 (N-R) 63.5(N-R) 2 Z s / d B A 70.2 72.1 1.9 Z n / d B A 60.4 65.3 • 4.9 SNA 1 dB 9.8 6.8 -3 E D T / s 0.47 0.50 0.03 STI (at # 5 receiver) 0.75 / Good 0.68 / Good -0.07 STI (at the other receivers) 0.06 0.04 -0.02 78 Table 9. Predicted acoustical values in different configurations with barriers. L A B / UC A B / U C L A B / A C A B / A C Lpm,n 1 dBA - voice range 71.8 (R-L) 63.9 (N-R) 63.5 (N-R) 62.2 (N-R) U /dBA 76.7 66.5 67.9 62.5 LJ dBA 76.1 67.4 65.3 60.9 SNA 1 dB 0.6 -0.9 2.6 1.6 EDT/s 0.96 0.56 0.55 0.21 STI (at # 5 receiver) / SI 0.45 / Poor 0.51 / Fair 0.54 / Fair 0.72 / Good STI (at the other receivers) / SP 0.05 0.06 0.04 0.02 In Fig. 36 the EDT values at different seating positions caused by a talker at seat 5 in two different occupancies, along with the averages, are shown. As expected, the EDT decreases significantly by adding absorption. With low absorption, EDT was almost constant across receiver positions. However, adding absorption caused the EDT to increase with increasing the distance from the talker. This effect is more noticeable in A C B configurations. EDT is another indicator of SI. By increasing EDT, SNA decreases; consequently SP would be preserved more. This demonstrates the effectiveness of barriers in preserving SP. 1.6 < S ^ ( V ( b * < b \ < b 9 > $ ^ t < y t < b ^ t < p A0> Receiver positions - • - 1 2 C - U C • 12C-AC - * ^ 1 2 C - A W 12C-AF - * - 1 2 C - A C B - • - 36C-UC + - 36C-AC - * - 36C-AW 36C-AF —•— 36C-ACB Figure 37: Distribution of predicted EDT along seats 0 to 15 caused by a talker at seat 5. 79 6.6 S u m m a r y Predictions with the CATT-Acoustic software, for a typical EE showed that adding absorption to EEs can increase the STI by 0.2 and, consequently, can increase SI from fair to good. For a given amount of absorption, adding absorption to the ceiling is more effective. Adding absorption to the walls had the same effect on SI and SP as adding absorption to the ceilings; however due to the larger surface area of the walls, the former treatment would be more expensive than the latter. Adding thin carpet to the floor didn't change SI significantly; however it improved SP. By comparing the results with low and high occupancies, decreasing the seating density or increasing the area per person wasn't as efficient as adding absorption. Moreover, adding absorption is less expensive than increasing the area per person. Inserting absorptive barriers in untreated configurations didn't improve the STI. The most effective treatment was inserting absorptive, 1.5 height barriers with the addition of absorptive ceilings, which led to good SI and normal voice level. Reducing the distance between talkers and receivers is another solution, but this may not always be feasible. Table size, receive-talker relationship and the number of the people in groups limit this solution. 80 Chapter 7 CONCLUSION 7.1 Research Contributions The evaluation of acoustical environments in EEs was performed in order to investigate how to optimize their acoustical characteristics and to design the optimal conditions for verbal communication. A review of the literature on speech intelligibility and speech privacy and their components in different contexts, such as open-plan offices and EEs was performed. It was found that although noise in EEs is an important factor that directly influences the health and well-being of workers and customers, there are a few studies on this subject. From the literature review on EEs, it was found that different seat densities or persons per unit area in EEs can affect the quality of verbal-communication the most. To evaluate the acoustical conditions in EEs, and the parameters that can affect them, ten typical EEs of four different types (restaurants, bistros, cafeterias and seniors' -residence dining rooms), on and off of the University of British Columbia campus, were studied. Physical measurements were made of Reverberation Time (RT) and noise levels in the unoccupied and occupied EEs. The noise exposure of the employees was also measured. Employees' daily noise exposures were calculated and compared with WorkSafeBC regulations. RTs were longer than optimal in most EEs. The noise levels measured were well above the recommended standards. However, these levels did not breach health-and-safety levels for employees; they were in compliance with WorkSafeBC occupational-noise regulations. To include the subjective evaluation of acoustical environments in EEs, two different versions (for customers and employees) of a questionnaire were developed and administered in the ten EEs. It was found that customers visit EEs for more reasons than just eating/drinking and talking. It was found that EE customers don't enjoy their visit i f there is inadequate speech intelligibility between people at their table, or inadequate speech privacy between tables. Lack of speech intelligibility at the table caused customers to suffer broken concentration and tired voices. The likely reason for the tired voices was the need to raise the voice to be heard over the 81 noise (Lombard effect). It was found that people talking and moving, and kitchen activity and equipment, were the most annoying sources of noise for employees. Hard-of-hearing employees reported less difficulty talking with colleagues and customers. However, the longer they had been hard-of-hearing, and the more severe it was, the more they needed to raise their voices to be heard, and the more they suffered broken concentration. Further analysis of measured noise levels was made to identify the relationship between noise level and occupancy (the number of customers) at a given time in EEs by taking into account the Lombard effect. The sound-power level, or voice level, of individual talkers in each dining period in each EE was calculated: it was found that, as noise levels increased in the EEs, customers needed to raise their voices to overcome the noise. The increase in voice level for 1 dB increase of noise was 0.7 dB. The "Lombard prediction" model was developed in parallel with this work to model speech and noise levels in an EE containing any number of customers/talkers, including the Lombard effect. The model contains a number of parameters known from measurements, and several unknown parameters. Optimization techniques were used to estimate the unknown parameters that fit the measured data the best. Talkers' voice levels with this model were used as inputs to predict speech intelligibility and privacy at a typical receiver position. Prediction was used to evaluate the speech conditions (speech intelligibility and privacy) at a typical receiver (seating) position in an EE. The CATT-Acoustic software was used, with the input talker' voice level from "Lombard prediction", to predict STI and its components, EDT and SNA. One of the studied EEs with simplified geometry was considered. In order to model the EE same as the actual one, some changes in materials were made to predict RTs same as measured. Predictions were made in different configurations, starting with the untreated configuration (original condition); then the materials used for the ceilings, floor and walls were changed to add some absorption. The effect of barriers of different heights and absorptions were also investigated. A l l predictions were made with low (25%) and high (75%) EE occupancies. It was found that, for a given amount of absorption, adding absorption to the ceiling is more effective. Adding absorption to the ceiling changed the STI from 0.41 in the untreated configuration to 0.55, with corresponding speech-intelligibility values of poor and fair, respectively. Adding absorption to the walls had the same effect as adding absorptive ceilings; however due to the larger surface area it would be more costly. Adding absorption to the floor 82 didn't change the STI. Inserting 1.5-m high, low- absorption barriers in the untreated configuration didn't change the STI significantly. Adding high-absorption barriers to the place increased STI from 0.41 to 0.51, with corresponding speech intelligibility values of poor and fair, respectively. The most effective treatment was inserting absorptive barriers with absorptive ceilings, which changed STI to 0.72, with corresponding speech intelligibility of good. Predictions with 1.2-m and 1.5-m-high barriers had almost the same results as with 1.5-m-high barriers. However, 1.7-m-high barriers are more costly, and the results with 1.2-m-high barrier may not be the same if the listener and receiver heights were greater than the 1 m assumed in this research. Comparing the results with low and high occupancies, it was found that enlarging the area per person (or decreasing the seating density) wasn't as effective as adding absorption. Decreasing the receiver to talker distance by half (from 1 m to 0.5 m) is another solution; however table size, the receiver-talker relationship and the number of people in groups limit the feasibility of these solutions. The main novelty of this work was including the Lombard effect in predicting talkers' voice levels in different noise levels and at different occupancies in EEs. This work was the first to measure the noise exposures of employees, along with soliciting information about the acoustical environment from their point of view in order to investigate health issues related to noise levels in EEs. 7.2 F u t u r e w o r k Further investigations could be done by simplifying and summarizing the questionnaires to include more subjects (customers and employees). Being present in the EEs at the time of measurements would help in recording the number of customers at different times, and avoid difficulties in interpreting questionnaires' responses. Including more seniors' homes would increase the number of hard of hearing subjects and result in better EE design with better acoustical environment for verbal communication. Performing predictions for more configurations, such as long geometry (high length-width ratio) would improve the applications of the results. 83 Bibliography 1. M. R. Hodgson and S. M. Kennedy, "Subjective assessments of listening environments in classrooms", J. Acoust. Soc. Am. 111(5, Pt. 2) (2002). 2. Bronkhorst, A. W., "The cocktail party phenomenon: a review of research on speech Intelligibility in Multiple Talker Conditions", ACUSTICA 86 (2000). 3. Cherry C. "Cocktail Party effect", Journal of the Acoustical Society of America 1953; 25: 975-979. 4. http://www.acoustics.com/ra restaurantnoise.asp. 5. Workers' Compensation Board of BC, " Occupational noise surveys", second edition, May 2001. 6. M. Hodgson. Course notes from Winter Term 2003 Mech 405: Acoustics and Noise Control; University of British Columbia. 7. A. Behar, E. MacDonald, J. Lee, J. Cui, H. Kunov and W. Wong, "Noise exposure of music teachers", Journal of Occupational and Environmental Hygiene, (1) 243-247 (April 2004). 8. W.C. Sabine, Collected Papers on Acoustics (Dover, New York, 1964). 9. J. S. Bradley, "Predictor of speech intelligibihty in rooms", Journal of Acoustical Society of America, 80(3), 837-845 (Sep 1986). 10. A. K. Nabelek and P.K. Robinson, "Monaural and Binaural Speech Perception in Reverberation for listeners of Various Ages," Journal of Acoustical Society of America, 71, 1242-1248 (1982). 11. ANSI S3.5-1997: Methods for the Calculation of the Speech Intelligibility Index (Acoustical Society of America, New York, 1997). 12. Sound System Engineering by Don and Carolyn Davis. 13. J.H. Rindel, "Computer Simulation Techniques for Acoustical Design of Rooms," Acoustics Australia 23, 81-86 (1995). 14. G.M. Naylor, "Treatment of Early and Late Reflections in a Hybrid Computer Model for Room Acoustics," 124th ASA Meeting, Paper 3aAA2 (New Orleans 1992). 15. B.I. Dalenback, CATT-Acoustic v8.0 (Gothenburg, 2004). 16. J. S. Bradley, "Speech intelligibility studies in classrooms," Journal of Acoustical Society of America, 80, 846-854 (1986). 17. T. Houtgast, H. J. Steenken and R. Plomp, "Predicting Speech Intelligibility in Rooms from the Modulation Transfer Function", ACUSTICA Vol. 46 (1980). 18. T. Houtgast and H. J. M. Steenken, "A multi-language evaluation of the RASTI-method for estimating speech intelligibility in auditoria," Acustica 54 (4), 185-199 (1984). 19. M. B. Gardner, "Factors affecting individual and group levels in verbal communication", Journal of the Audio Engineering Society 19(7) 560-569 (1971). 84 20. J. A . Veitch, J.S. Bradley, L . M . Legault, S.G. Norcross and J.M. Svec, "Masking speech in open-plan offices with simulated ventilation noise: noise level and spectral composition effects in acoustic satisfaction", IRC report (2001). 21. J.S. Bradley, "Acoustical design of open-plan offices", IRC report (October 2004). 22. Petra Larm, Jukka Keranen, Riikka Helenius, Jarkko Hakala, Valtteri Hongisto, "Acoustics in open-plan offices-a laboratory study", Joint Baltic-Nordic Acoustic Meeting 2004, 8-10 June 2004, Mariehamn, Aland. 23. R:" Moulder, "Quiet Areas in Restaurants", Report by Battelle to the U . S. Architectural and Transportation Barriers Compliance Board, 1993. 24. A . White, "The effect of the building environment on occupants: the acoustics of dining spaces", M.Phil. Dissertation, University of Cambridge, 1999. 25. A. Astolfi and M . Filippi, "Good acoustical quality in restaurants: a comparison between speech intelligibility and privacy", Proc. EuroNoise 2003, Naples, Italy, SI02 (2003). 26. J. Kang, "Numerical modelling of the speech intelligibility in dining spaces", Applied Acoustics 63(12) 1315-1333 (2002). 27. L . H . Christie, "Psycho-to-building acoustics: are bars, cafes, and restaurants acceptable acoustic environments?", Research Report, Victoria University of Wellington, 2004. 28. E. Lombard, "Le signe de l'elevation de la voix" Annales des Maladies de l'Oreille, du Larynx, du Nez et du Pharynx 37, 101-119 (1911). 29. T. S. Korn, "Effect of psychological feedback on conversational noise reduction in rooms", Journal of Acoustical Society of America 26(5) 793-795 (1954). 30. J. M . Pickett, "Limits of direct speech communication in noise", Journal of Acoustical Society of America 30(4) 278-281 (1958). 31. S. K. Tang, D. W. T. Chan, and K. C. Chan, "Prediction of sound-pressure level in an occupied enclosure", Journal of Acoustical Society of America 101(5) 2990-2993 (1997). 32. G. Dodd and J. Whitlock, "Auditory and behavioural mechanism influencing speech intelligibility in primary school children", Proc. 18th International Congress on Acoustics, Kyoto, 3581-3582 (2004). 33. H . Sato and J. S. Bradley, "Evaluation of acoustical conditions for speech communication in active elementary school classrooms", Proc. 18 th International Congress on Acoustics, Kyoto II, 1187-1190 (2004). 34. L. Sutherland, D. Lubman and K. Pearsons, "Acoustic environment challenges for the unique communication conditions in group learning classes in elementary school classrooms", Journal of Acoustical Society of America 117(4, Pt. 2) 2366 (2005). 35. Wayne O. Olsen, "Average Speech Levels and Spectra in Various Speaking/Listening Conditions: A Summary of the Pearson, Bennett & Fidell (1977) Report", American Journal of Audiology, 7(2), (December 1998). 36. Workers' Compensation Board of B.C., "Proposed OHS Guidelines - Part 7: Noise Exposure" (2004). 85 37. A. Behar, E. MacDonald, J. Lee, J. Cui, H. Kunov and W. Wong, "Noise exposure of music teachers", Journal of Occupational and Environmental Hygiene, (1) 243-247 (April 2004). 38. M . Hodgson, G. Steininger, Z. Razavi, "Measurement and Prediction of Speech and Noise Levels and the Lombard Effect in Eating Establishments," Submitted to Journal of Acoustical Society of America, September 2006. 39. F. Lawson, Catering design. In: Tutt T, Adler D, editors. New Metric Handbook. London: The Architectural Press Ltd, 1979. 40. J. S. Bradley, "The Acoustical design of conventional open plan offices", Canadian Acoustics, Vol. 27, No 3(2003), 23-30. 86 APPENDIX A EE Photographs and Floorplans Windows Windows Eating area Eating area Open to SUB 0 L 2 JL 4 J _ 6 J -8 I 1 0 m (C2) > O pen to S U B E ating area O p e n to S U B 0*3) 89 91 (Rl) • Entrance r- o 0 0 0, oo a OTTTJH [TJ-Tjrj 0 0 0' o C P 0 0 0 0 0 0 0 0 0 • n BS 0 o Q > Ear h 4 h 6 h 8 •— 10 m 93 94 95 O'Keefe Residence (SI) Kilchcn Coffee bar E a l i n g a r e a Windows 2 -L 4 8 10m -I 96 APPENDIX C - Consent Form T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A School of Occupational & Environmental Hygiene 3 r d Floor- 2206 East Mall Vancouver, BC Canada V6T 1Z3 www.soeh.ubc.ca (604) 822-9595 te/ (604) 822-9588 tax Consent Form Acoustical Evaluation and Improvement of Eating Establishments Principal investigator: Murray Hodgson, Ph.D., Professor Occupational and Environmental Hygiene, UBC Phone number: 604- 822-3073 Fax number: 604-822-9588 Email: hodgson@mech.ubc.ca Researcher: Zohreh Razavi Master of Applied Science candidate Mechanical Engineering Department, UBC Phone: 604-822-9575 Email: razaviz@mech.ubc,ca Dear Restaurant Employee, The purpose of this study is to determine the most desirable sound environments for workers and customers in different eating establishments. The study will involve physical measurements of noise exposure, and questionnaires for workers and customers. This is a graduate research project and we would like to invite you to participate. Physical measurements for objective evaluation will provide us with information about sources of noise and their characteristics. In order to take the noise measurements, we would like to ask you to wear a device for monitoring noise level on a designated workday. This device weighs less than 2 pounds and is attached to a belt and worn around the waist. A small microphone attached to the meter by way of a wire will be clipped to your collar. This device should not interfere with any of your job duties. This instrument DOES NOT record or transmit conversations; dialogue or speech cannot be retrieved from the data collected - it only monitors the volume of the sound. Attachment and removal of the device by the researcher will take approximately 5 minutes at the beginning and end of your shift. At the end of your shift we may follow-up with some questions about your workday. Please treat your workday as any normal workday when you are wearing one of these noise meter devices. This device is meant to be worn by only you for the full day regardless of what you are doing. Version: Jan 07, 2005 Page 1 of 2 9 8 We expect that the results of this study will help to identify possible modifications to the various eating establishments studied to make them more accessible for normal-hearing and hearing-impaired individuals, and allow for more effective verbal communication between workers and customers or other workers, as well as to improve the dining experience. Your participation in this survey is completely voluntary. You may choose to withdraw at any time without prejudice to your present employment. The principal investigator and the researcher will be pleased to answer any questions or concerns you may have about this study. Please feel free to contact them if you have any questions. If you have any concerns about your treatment or rights as a research subject, you may telephone the Research Subject Information line in the UBC Office of Research Services at the University of British Columbia at 604-822-8598. Your signature below indicates that you have read and received a copy of this consent form for your own records. Your signature also indicates that you consent to participate in this study. I consent/ I do not consent (circle one) to my participation in this study. Your printed name: Your signature: Date: Version: Jan07, 2005 Page 2 of 2 99 APPENDIX D - Employee Questionnaire T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A School of Occupational & Environmental Hygiene 3 , d Floor - 2206 East Mall Vancouver, BC Canada V6T 1Z3 www.soeh.ubc.ca (604) 822-9595 tel (604) 822-9588 fax A c o u s t i c a l Eva lua t i on and Improvement in Ea t i ng Es tab l ishments Dea r emp loyee , W e are s tudy ing the acoust ica l (sound) env i ronments in eat ing establ ishments at U B C . T h i s study w i l l i nvo l ve tak ing phys ica l measurements o f no ise , a l ong w i th data co l lec ted through se l f -admin is tered quest ionnaires. T h i s is a graduate research project and we w o u l d l ike to invi te y o u to part ic ipate. P h y s i c a l measurements in the eat ing establ ishments w i l l p rov ide us w i th in fo rmat ion about sources o f noise and their character is t ics. The quest ionnai re w i l l help us to evaluate the sound env i ronment in the eat ing estab l ishment f rom you r point o f v i e w . W e expect that the results o f this study w i l l help to ident i fy poss ib le mod i f i ca t i ons to the eat ing establ ishments studied to make them more access ib le for no rma l -hear ing and hear ing- impa i red ind iv idua ls . These mod i f i ca t ions w i l l make the establ ishments more ef fect ive for verbal commun i ca t i on between worke rs and customers or other worke rs . T h e quest ionnaire shou ld take no longer than 10 minutes to comple te . Par t ic ipa t ion in the survey is comp le te l y vo lun tary and a l l quest ionnai res are anonymous . C o m p l e t e d quest ionnaires w i l l be kept str ic t ly con f iden t ia l . I f the quest ionnai re is comp le ted and returned to us, it is assumed that consent to part ic ipate in this study has been granted. T h e pr inc ipa l invest igator and the researcher w i l l be pleased to answer any quest ions or concerns y o u may have about this study. P lease feel free to contact them i f y o u have any quest ions. Y o u r name w i l l not appear anywhere on the comp le ted quest ionnai re . If y o u w o u l d l ike to receive a copy o f a summary o f the results o f this s tudy, please contact us d i rec t ly at the ema i l addresses noted be low. I f y o u have any concerns about you r treatment or r ights as a research subject, y o u may te lephone the Research Subject In format ion l ine in the U B C O f f i c e o f Research Serv ices at the Un ive rs i t y o f Br i t i sh C o l u m b i a , at 604 -822 -8598 . T h i s is an important study. Thanks for you r va luab le cont r ibut ion to it. Bes t regards, Principal Investigator: Murray Hodgson, P h . D . , Pro fessor , Occupa t i ona l and Env i r onmen ta l H y g i e n e , U B C , phone: 822 -3073 , e m a i l : h o d g s o n @ m e c h . u b c . c a Researcher: Zohreh Razavi, Mas te r o f A p p l i e d Sc ience candidate, M e c h a n i c a l Eng inee r i ng Depar tment , U B C , phone: 822 -9575 , e m a i l : r a z a v i z @ m e c h . u b c . c a V e r s i o n : Jan 07 , 2005 Page 1 o f 4 100 Evaluation of Sound Environments in Eating Establishments - Employees The purpose of this questionnaire is to allow you, the employee, to evaluate the sound environment in this eating establishment. It consists of two parts - noise evaluation and general information. Please respond to all questions about this eating establishment from your point of view on this day, as accurately as possible. The questionnaire is completely anonymous. Thank you for your kind cooperation. Date: Name of the eating establishment: Beginning of your shift: am • , pm • End of your shift: am • , pm • What is your job title? -Noise Evaluation 1. Please indicate to what extent each of the following sources of noise bothers you while you are working in this eating establishment. Not at all A little Some A lot Very much > Customers' talking • • • • • > Customers' movement • • • • • > Colleagues' talking • • • • • > Colleagues' movement • • • • • > Cel l phone ringing or conversation • • • • • > Activities in the kitchen • • • • • > Kitchen equipment (fridge, coffee maker,....) • • • • • > Heating/Ventilation/Air conditioning • • • • • > Noise from outside the eating establishment • • • • • > Music • • • • • > Other (specify ) • • • • • 2. Please indicate to what extent the noise in t your experiencing the following problems: l is eating establis linent contributes to None A little Some A lot Very much > Difficulty in talking with colleagues • •• • • • > Difficulty in talking with customers • • • • • > Broken concentration • • • • • > Frustration • • • • • > Need to raise your voice to be heard • • • • • > Headache • • • • • > Fatigue • • • • • > Tinnitus (ringing in your ears) • • • • • > Other (specify ) • • • • • Version: Jan07, 2005 Page 2 of 4 101 3. Please indicate how well the acoustical environment in this eating establishment today met your expectations for an ideal work environment. Very well • Well • Average • Poorly • Very poorly • 4. Please indicate to what extent you can easily hear the following conversations while you work and are not a part of their conversations. Not at A Some A Very all little f lot much > Conversations at the tables 0 0 0 0 0 > Conversations between your coworkers 0 0 0 0 0 > Other (specify ) 0 0 0 0 0 B. General information > Sex: Female 0 Male 0 > Age: ! > Is English your first language? Yes 0 No 0 > Is English the main language spoken in this eating establishment? Yes 0 No 0 > Are you aware of having a hearing impairment? Yes 0 No 0 > If yes, how severe? (please estimate if not sure) Mild Moderate Moderately severe Severe Profound O O O 0 0 > For how many years or months have you had this hearing problem? > For how many years have you been working in this eating establishment? ^ For how many hours per week do you work in this eating establishment? Version: Jan 07, 2005 Page 3 of 4 102 > Please specify your working area/areas you spend the majority of your time by circling the appropriate numbers. Please deposit this questionnaire in the provided box. Thank you and have a great day. Version: Jan07, 2005 Page 4 of 4 103 APPENDIX E - Customer Questionnaire T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A School of Occupational & Environmental Hygiene 3 , d Floor- 2206 East Mall Vancouver, BC Canada V6T 1Z3 www.soeh.ubc.ca (604) 822-9595 te/ (604) 822-9588 fax Acoustical Evaluation and Optimization in Eating Establishments Dear Restaurant Customer, We are studying the acoustical (sound) conditions in eating establishments at U B C . This study wi l l involve physical measurements of sound propagation along with data collected through self-administered questionnaires. This is a graduate research project and we would like to invite you to participate. Physical measurements for objective evaluation wi l l provide us with information about noise sources, distribution, and power. The data necessary to evaluate the acoustical conditions in the eating areas from the point of view of customers wi l l be the focus of the questionnaires. We expect that the results of this study wi l l help to identify possible modifications to the various eating areas studied to make the establishments more accessible for normal-hearing and hearing-impaired individuals and more effective for verbal communication between workers and customers. The questionnaire should take no longer than 10 minutes to complete. Participation in the survey is completely voluntary and all questionnaires are anonymous. Completed questionnaires wi l l be kept strictly confidential. I f the questionnaire is completed and returned to us, it is assumed that consent to participate in this study has been granted. The principal investigator and the researcher wi l l be pleased to answer any questions or concerns you may have about this study. Please feel free to contact them i f you have any questions. Your name wi l l not appear anywhere on the completed questionnaire. If you would like to receive a copy of a summary of the results of this study, please contact us directly at the email addresses noted below. If you have any concerns about your treatment or rights as a research subject, you may telephone the Research Subject Information line in the U B C Office o f Research Services at the University of British Columbia, at 604-822-8598. Best regards, Principal investigator: Murray Hodgson, Ph.D., Professor, Occupational and Environmental Hygiene, U B C , phone: 822-3073, email: hodgson@mech.ubc.ca Researcher: Zohreh Razavi, Master of Applied Science candidate, Mechanical Engineering Department, phone: 822-9575, email: razaviz@mech.ubc.ca Version: Jan 07, 2005 Page 1 of 5 104 Evaluation of Sound Environments in Eating Establishments - Customers The purpose of this questionnaire is to allow you, the customer, to evaluate the sound environment in this eating establishment and help us to learn how to make this establishment more desirable for customers and workers. It consists of three parts - your reasons for coming to this eating establishment, noise evaluation and general information. Please respond to all the questions, from your point o f view, about this eating establishment, at this time, as accurately as possible. The questionnaire is completely anonymous. Thank you for your kind cooperation. Date: Name o f the eating establishment: Arr ival time: am • , pm • Departure time: am • , pm • A. Your reasonts) for coming to this eating establishment What were your reason (s) for coming to this eating establishment and with respect to those reasons you have checked, indicate how well the sound environment in this eating establishment met your expectations for an ideal environment Please check all that apply. Very well Wel l Average Poorly Very Poorly • Eating/drinking (coffee,beer...) • • • • • • Talking to friends/coworkers • • • • • • Business meeting • • • • • • Studying, working • • • • • • Celebration, entertainment • • • • • • Relaxing • • • • • • Other (specify ) • • • • • ^ How frequently do you come to this eating establishment? For the first time • More than once a month • Less than once a month • Almost every day • ^ Do you prefer any specific seating location in this eating establishment? Yes O N o 0 ^ If yes, is this because of noise concerns in this eating establishment? > Yes • N o D ^ If yes, please describe Version: Jan 07, 2005 Page 2 of 5 105 ^ H o w many people ( i nc lud ing you ) were hav ing this mea l together? Please spec i fy the number o f people in each o f the f o l l o w i n g groups: ch i ld ren adults seniors H o w many customers were in the restaurant du r ing you r v i s i t? ( i f not sure, p lease guess) B. Noise Evaluation 1. P lease indicate h o w much each o f the f o l l o w i n g sources o f no ise in this eat ing estab l ishment bothers y o u du r ing this v is i t : N o t at a l l A l i t t le S o m e A lot V e r y m u c h > Peop le speak ing • • • • • > Peop le m o v i n g around • • • • • > Ad jacen t table conversat ion • • • • • > C e l l phones • • • • • > H e a t i n g / A i r cond i t i on ing /Ven t i l a t i on • • • • • > A c t i v i t i e s in the k i tchen • • • • • K i t c h e n equ ipment ( f r idge, co f feemaker , etc.) • • • • > C l i n k i n g sounds o f d ishes, cut lery , etc.) • • • • • > N o i s e f r om outs ide the eat ing estab l ishment • • • • • > M u s i c • • • • • > Reverbera t ion (echo) • • • • • > Othe r (spec i fy ) • • • • • V e r s i o n : Jan 07, 2005 Page 3 o f 5 106 2. Please indicate to what extent no ise in this eat ing establ ishment contr ibutes to you r exper ienc ing the f o l l o w i n g p rob lems du r i ng this v is i t : N o t at a l l A l i tt le S o m e A lot V e r y m u c h > Reduced en joyment o f eat ing out • • • • • > D i f f i c u l t y in hear ing conversa t ion at y o u r table • • • • • > D i f f i c u l t y in ta l k ing w i th wai ter /wai t ress • • • • • > B r o k e n concent ra t ion • • . • • • > T i r e d vo i ce • • • • • > Headache • • • • • > Fat igue • • • • • > Other (spec i fy ) • • • • • 3. Please indicate to what extent y o u can eas i l y hear conversat ions f r om adjacent tables: N o t at a l l • A l itt le • S o m e • A lot • V e r y much • 4 . P lease indicate to what extent y o u feel others can overhear conversat ions at you r table: N o t at a l l • A litt le • S o m e • A lot • V e r y much • C. General information > Sex : Fema le • , M a l e • > A g e : > Is E n g l i s h you r f i rst language? Y e s • N o • > Is E n g l i s h the ma in language spoken in this eat ing estab l ishment? Y e s • N o • \ V e r s i o n : Jan 07, 2005 Page 4 o f 5 107 ^ D u r i n g this v is i t , w o u l d y o u prefer to have you r meal in a no isy or quiet sett ing? N o i s y • Qu ie t • Doesn ' t matter • ^ D u r i n g this v is i t , w o u l d y o u prefer to have mus ic w i th you r mea l? Y e s • N o • D o e s n ' t matter • If yes , what leve l o f mus i c w o u l d y o u prefer? V e r y h ighD H i g h D Mode ra teD L o w D V e r y L o w • ^ A r e y o u aware o f hav ing a hear ing impa i rment? Y e s • N o • > I f yes , h o w severe? (please est imate i f unsure): M i l d Modera te Modera te l y severe Severe P r o f o u n d • • • • • > Please use this space to tel l us about any important issues about no ise in this eat ing estab l ishment that w e may have m issed , or any other comments y o u may have : ^ P lease mark w i th * you r approx imate seat ing pos i t ion in the gray area, this t ime, in this eat ing establ ishment. P lease deposi t th is quest ionnai re in the prov ided box. Thank y o u and have a great day. V e r s i o n : Jan 04, 2005 Page 5 o f 5 108 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0080713/manifest

Comment

Related Items