Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Towards an objective measure of speakers' intelligibility derived from the speech wave envelope Hoek, Dorothy Christine 1988

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

Item Metadata

Download

Media
831-UBC_1988_A6_7 H63.pdf [ 4.41MB ]
Metadata
JSON: 831-1.0097698.json
JSON-LD: 831-1.0097698-ld.json
RDF/XML (Pretty): 831-1.0097698-rdf.xml
RDF/JSON: 831-1.0097698-rdf.json
Turtle: 831-1.0097698-turtle.txt
N-Triples: 831-1.0097698-rdf-ntriples.txt
Original Record: 831-1.0097698-source.json
Full Text
831-1.0097698-fulltext.txt
Citation
831-1.0097698.ris

Full Text

TOWARDS AN OBJECTIVE MEASURE OF SPEAKERS'  INTELLIGIBILITY  DERIVED FROM THE SPEECH WAVE ENVELOPE  by  DOROTHY C . HOEK B.A.,  UNIVERSITY OF BRITISH COLUMBIA,  1985  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE SCHOOL OF AUDIOLOGY AND SPEECH SCIENCES THE FACULTY OF MEDICINE  We accept t h i s t h e s i s  as  conforming  to the r e q u i r e d standard  THE UNIVERSITY OF BRITISH COLUMBIA August, ®Dorothy  1988  C h r i s t i n e Hoek,  1988  In presenting  this thesis in partial fulfilment  of the requirements for an advanced  degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department  or  by  his  or  her  representatives.  It  is  understood  that  copying  or  publication of this thesis for financial gain shall not be allowed without my written permission.  Department The University of British Columbia 1956 Main Mall Vancouver, Canada V6T 1Y3  DE-6G/81)  ABSTRACT  This  study  investigates  r e l a t i o n s h i p between envelope  and  clarity. called  French  samples  speaker  A computer based  on  of  perceptual weak  affected the  speakers intended  or  the  a  speech  articulatory measure  English  and d i g i t i z e d .  listeners  conditions:  calculation  modulation  speech  sample.  of  depth  judged  MI,  i n the  The MI v a l u e s s o  the corresponding r a t i n g s who  were  and O v e r a r t i c u l a t e d .  for  of amplitude  a n d one  Speakers  three a r t i c u l a t o r y  developed  compared w i t h  speakers  the  from  articulatory  of the recorded u t t e r a n c e s .  Results  and  under  each d i g i t i z e d  English-speaking  is  several  Normally A r t i c u l a t e d ,  p r o g r a m was  o b t a i n e d were  clarity  from  t h e amount  envelope  in  of  Index ( M I ) .  speech  Underarticulated,  modulation  intelligibility  were r e c o r d e d  t o produce  possibility  a t d e v e l o p i n g an i n t e l l i g i b i l i t y  the Modulation  Speech  asked  amplitude  a speaker's  I t aims  the  indicate  that  the  relationship  d a t a and t h e M o d u l a t i o n  and non-monotonic. the r e s u l t s  of the comparison data.  were  always  not  articulatory there  the  i n i t s present  form  Several  perceptual  precautions,  Index  between  There  factors  are  successful  differences  d u r a t i o n between u t t e r a n c e s f r o m  have  between t h e MI v a l u e s  conditions.  were some  may  the three  indications in  producing Also,  in  that the  despite  i n t e n s i t y and  conditions.  It  i s concluded  amplitude  modulation  intelligibility Modulation become  a  explored, Modulation for  that  or  Index useful and  in  will  require  tool.  and  speech  articulatory  possible  Index  t h e r e i s some c o r r e l a t i o n between  Some further  envelopes clarity.  modification modifications modifications  the e x p e r i m e n t a l  future investigations.  design  and  speakers'  However,  the  before  i t can  were  briefly  to are  both  the  suggested  iv  TABLE OF CONTENTS  CHAPTER  PAGE  ABSTRACT  i i  TABLE OF CONTENTS  iv  L I S T OF TABLES  v i i  L I S T OF FIGURES  .  ix  L I S T OF APPENDICES  xi  ACKNOWLEDGMENT  x i i  1.  INTRODUCTION  1  2.  LITERATURE REVIEW  6  2.1  Introduction  2.2  Factors  6  A f f e c t i n g Speech  Intelligibility  2.21  Factors  i n v o l v i n g the speaker  2.22  Factors  involving the transmission  2.3  Speech  Intelligibility  Listener  Measures  6 system.  13  Intelligibility  2.32  D e t a i l e d judgment-based  scales..  13 intelligibility  test Acoustic  11  B a s e d on  Judgments  2.31  2.4  6  16 Intelligibility  Measures  17  (AI)  17  2.41  The A r t i c u l a t i o n  2.42  The S p e e c h T r a n s m i s s i o n  2.43  Modified  2.44  Articulation  Speech  Index  Index  Transmission  ( S T I , RASTI) Index  Loss o f Consonants  (mSTI).  (ALcons)..  21 27 27  2.45  Direct-to-Reverberant  Intensity  Method  (SRR)..... 2.46  Speech Communication  2.47  Pattern  2.48  K o n d r a s k e ' s Method  31  2.49  Monsen's F o r m u l a  31  Comparisons  2.6  F u r t h e r Developments  2.7  Index  (SCI)  C o r r e s p o n d a n c e Index  2.5  Transfer  3.  28  of Acoustic  30  (PCI)  30  Indices  32  of the Modulation  Function  33  Conclusion  METHODS AND  33  MATERIALS  35  3.1  An O v e r v i e w o f t h e E x p e r i m e n t a l D e s i g n  3.2  Preparation  o f Speech  Samples  35 ;  36  3.21  Speech m a t e r i a l  36  3.22  Speakers  36  3.23  R e c o r d i n g o f speech samples  37  3.3  Design of the L i s t e n i n g  Test  3.31  Preparation  3.32  Listeners  3.33  Procedures f o r the L i s t e n i n g  3.4  of the L i s t e n i n g  39 Test  39 43  Test  The M o d u l a t i o n Index  3.41  Description  3.42  Digitization  43 44  i f t h e Index  44  o f speech samples  4. RESULTS 4.1  tape....  46 51  Results  of the L i s t e n i n g  4.11  Consistency  4.12  Comparison  Test  of l i s t e n e r of l i s t e n e r  judgments judgments  51 51  vi  with  speaker i n t e n t i o n s  55  4.13  Effect  o f l a n g u a g e on l i s t e n e r  4.14  Effect  of utterance duration v a r i a b i l i t y  on l i s t e n e r  judgments..  judgments  60  4.2  R e s u l t s o f t h e MI C a l c u l a t i o n s  4.3  Comparison  Comparison with  4.32  5.  visual  64  o f M o d u l a t i o n Index V a l u e s  listener  Comparison with  60  o f M o d u l a t i o n Index V a l u e s  with P e r c e p t u a l Data 4.31  59  judgments  64  o f M o d u l a t i o n Index V a l u e s identification  DISCUSSION AND CONCLUSIONS  o f waveforms...  70 75  BIBLIOGRAPHY  81  APPENDICES  86  vii L I S T OF  TABLES  TABLE  PAGE  I.  Information  II.  Duration  r e g a r d i n g the  speakers  i n seconds o f the  selected...  utterances  selected  42  III.  Information  IV.  Standard  V.  by  speakers  Standard  judgments  f o r Speakers 0 to 6 i n c l u s i v e .  ten l i s t e n e r s  a c r o s s the  i n t e n d e d by  f u n c t i o n of the  articulatory  the speakers  53 as  speaker  R e p e a t a b i l i t y of l i s t e n e r  54 judgments  across  listeners VIII.  Modulation six  IX.  54 Index v a l u e s  Confusion  speaker  of utterances  matrix  comparing  63 visual  of a r t i c u l a t o r y  c o n d i t i o n with  intentions  Confusion  by  condition  identification  XI.  by 62  Index r a n k i n g s  articulatory X.  for utterances  speakers  Modulation  53  judgments  R e p e a t a b i l i t y o f l i s t e n e r judgments a  44  conditions intended  deviations for listener  conditions  VII.  listeners  deviations for listener  the a r t i c u l a t o r y  by  VI.  r e g a r d i n g the  across the  40  matrix  identification  comparing  71 visual  of a r t i c u l a t o r y c o n d i t i o n  v i i i  with  the a r t i c u l a t o r y  condition  by t h e p e r c e p t u a l r a n k i n g to XII.  according  l i s t e n e r judgments  Comparison original to  order  suggested  MImod  MI  of rankings values with  71 according to the rankings  according 74  ix  L I S T OF  FIGURES  FIGURE  1.  PAGE  Illustration of  of the r e d u c t i o n  a speech s i g n a l  c a u s e d by  i n modulation background  n o i s e and r e v e r b e r a t i o n 2.  The N a t i o n a l Deaf  Technical  4 Institute  f o r the  (NTID) S c a l e o f I n t e l l i g i b i l i t y  3.  A percent  rating  4.  R e l a t i o n between  scale  14  of i n t e l l i g i b i l i t y . . . .  Articulation  Index  s c o r e s and PB-word s c o r e s 5.  The way  i n which  Transmission of  20  the o r i g i n a l  Speech  Index a c c o u n t e d f o r t h e  effects  background n o i s e  23  6.  R e l a t i o n between  7.  Labelling  o f p e a k s and  amplitude  e n v e l o p e o f a s p e e c h sample  8.  Speech  9.  An  sample  example  chosen  STI and PB-word s c o r e s  45  scheme  a n d end p o i n t  47 locations  f o r M o d u l a t i o n Index a n a l y s i s  of the  a m p l i t u d e e n v e l o p e o f a s p e e c h sample 10. L i s t e n e r  judgments  as a f u n c t i o n  56 judgments  as a f u n c t i o n  of the  listener 12. The  49  of the  speaker 11. L i s t e n e r  26  troughs i n the  digitization  of s t a r t  15  relationship  57 between  M o d u l a t i o n Index  v a l u e s and l i s t e n e r  judgments 12a  66  12b  67  12c.  68  12d  69  xi  LIST  OF APPENDICES  APPENDIX  A.  PAGE  Intelligibility  tests  based  on l i s t e n e r  judgments  86  B.  The E n g l i s h  C.  Listening and  D.  and F r e n c h  test  sentences  selected...  instructions  answer s h e e t  L i s t i n g o f t h e program the Modulation  88  Index  89 f o r calculation of 91  x i i  ACKNOWLEDGEMENT  I would thesis.  l i k e t o t h a n k e v e r y o n e who h a s h a d a p a r t In p a r t i c u l a r , I would l i k e t o thank:  i n this  Dr. A n d r e - P i e r r e Benguerel f o rsharing a small part of his knowledge with me, and f o r h i s assistance throughout the preparation o f t h i s t h e s i s . N o e l l e Lamb f o r h e r e n c o u r a g e m e n t s e r v i n g on my c o m m i t t e e . My s u b j e c t s  f o rtheir  My f a m i l y a n d M i k e My c l a s s m a t e s , friendship.  kind  and f o r  cooperation.  f o rtheir  especially  and i n t e r e s t ,  patience  and s u p p o r t .  Charlotte, f o r their  1 CHAPTER  ONE  INTRODUCTION  Intelligibility speech  may be  which  o f speech  u n d e r s t o o d by normal  i s unintelligible  conditions,  such as  been u n i n t e l l i g i b l e Various measure NTID  as i t l e f t have  source the  for  detail  listener  large 1987).  Furthermore,  Index,  speech  the degradation  Two, b u t  the present  subjective  and  intelligibility while others,  are  based  of  signals  a t the  including  intelligibility  measures,  will  some g e n e r a l  study are i n order.  be  at  over  the  speech  of  listeners  sample g r o u p s , addition,  affect  their  and i n c r e a s e d  Methods  judgments,  analysis  while l i s t e n e r s  discussed  observations  judgments have v a r i o u s d i s a d v a n t a g e s .  In  such as t h e  some methods were d e s i g n e d  assume f u l l  intelligibility  biases  to  systems.  i n Chapter  motivated  attempts  a r e o b j e c t i v e measures  1985, f o r e x a m p l e ) ,  a n d measure  Speech  in  of these,  are  have  lips.  developed Some  1947),  measurement o f  Articulation  transmission  and  analyses.  (Kondraske,  source  been  i t may  O t h e r s , such as t h e A r t i c u l a t i o n  (French & Steinberg,  adapted  the speaker's  or  interfering  ratings.  Index  or  noise,  Speech  1987),  on l i s t e n e r  acoustic  or  by  which  i n Doyle,  based  on  t h e ease w i t h  may have been a f f e c t e d  intelligibility.  (cited  to  hearing listeners.  reverberation  methods  speech  scale  refers  time  in  which  involving  The e x p e r i e n c e necessitating ( c f .  may make g r o s s  Doyle,  judgments  about  the adequacy  acoustic  analyses  attributes  to  remedy t h e  Without  reverberation,  engineers  tiles,  on t h e  or  other  transmission  which  speech  from  consuming physical  identify  the  particular  can  to score,  factors  auditorium  are  perceptual  qualities  is  o f poor known  speech to  give  to the  problem.  to  speech  unknown, t h e s o l u t i o n t o and e r r o r .  Even  affecting  contrast,  contributing  to  those  consistent  are  information the  those  individual  give  measurement, some  be  by p l a c i n g  the acoustic c o r r e l a t e s of  to  the  be b a s e d on  detrimental  errors  indices, i n  and  a t best  means s p e c i f i c  the  of  e r r o r s , attempts  improve t h e s i t u a t i o n  by o t h e r  measurement  aspects  i f the source  auditorium  not reveal  Acoustic  ratings  the  perceptual  o n l y be f o u n d by t r i a l  s o u n d s do  errors.  an  hand,  i n that  problem can  tests  knowledge o f  F o r example, in  the  pinpointing  are behind  intelligibility  If,  c a n n o t make t h e f i n e  problems i n t h e system w i l l  guesses.  acoustic  for  they  o f a speech s i g n a l which c o n t r i b u t e t o i t s r e l a t i v e  s i g n a l which  educated  speech,  necessary  intelligibility. acoustic  of the  less as  time  to  the  intelligibility  of  speech. The possible the  purpose o f  Modulation Transfer  use o f  speech  study  was  a p p l i c a t i o n o f the concept behind  intelligibility the  the present  a t the  t h e MTF  transmission  Function source,  (MTF)  to  -  to  an  i . e . the speaker.  or to  the  an a c o u s t i c i n d e x  c o n c e p t has been c o n f i n e d systems,  explore  index  of  Previously,  to the study o f  psychoacoustic  perceptual  3 studies. of  The MTF r e f l e c t s  an a c o u s t i c  the  wave p r o p a g a t i n g  intervening  noise, for  transmission  and f r e q u e n c y  an i l l u s t r a t i o n  Houtgast from  measures  obtained using  systems  correlation  authors  proceeded  to  Houtgast (1973)  found  a  intelligibility  near  s c o r e s and  in  the r e c e i v e d  of  modulation of  speech  compared clarity  tool  measures,  the  MTF a s an i n t e g r a l  part  Houtgast  relationship of  and Steeneken between  amplitude  a s m e a s u r e d by t h e MTF. signal  study, the  f o r assessment  articulatory of modulation to l i s t e n e r ' s  seems  to  speech  modulation Thus,  depth  correlate  with  might  were made affect  g o a l was t h e d e v e l o p m e n t  of i n t e l l i g i b i l i t y  clarity. depth  The  as a f f e c t e d  scores obtained  ( t h e M o d u l a t i o n Index  perceptual  f o r r e c o r d e d speech  Efforts which  t h e two  scores.  ( S t e e n e k e n & H o u t g a s t 1980;  retention  a speech  the present  speaker's measure  word  intelligibility.  In rating  signal  obtained  intelligibility  Balanced  1971, 1985). linear  reduction).  values  speech  the  T r a n s m i s s i o n Index  and Steeneken  to  between  incorporate  (see Figure 1  of modulation MTF  when  reverberation,  contaminants  Phonetically  a strong  depth  source to r e c e i v e r  (1973) compared  transmission  t h e Speech  as  i n modulation  introduces  of the p r i n c i p l e  Finding  of  from  system  filtering  and Steeneken  various  the reduction  samples  judgments from  of  several  of a by a  from a  o r MI) were articulatory  speakers.  to  minimize  contaminating  the  results.  Background  variables noise  and  Transmitted speech signal  Received speech signal modulation index «= m < 1  &S0703  F i g u r e 1,  I l l u s t r a t i o n of the r e d u c t i o n i n m o d u l a t i o n o f a speech s i g n a l caused by background n o i s e and r e v e r b e r a t i o n , (reproduced from Steeneken and Houtgast, 1985).  5 reverberation,  which  could  amplitude  modulation  in  minimized  by making  a l l of  with acoustic  tiling  booth quality  recording  filtering,  and  quality.  (MI)  effect  on  duration  the  intensity  and  received,  feedback  experimenter, during  avoid  words o r  to  predict.  speakers  as c o n s t a n t  to  their  were u r g e d  High  frequency  the r e c o r d i n g  differences,  spacing of  of speech  they  to  proof  ceiling.  absolute i n t e n s i t i e s ,  difficult  as  and  sound  present formulation,  timing  in  in a  were  the  new  and  the  whether  in  duration  of  Therefore, in  practiced  keeping  as p o s s i b l e ;  performance  from  to maintain t h i s  they the  constancy  recording sessions.  These influence depth  p r e c a u t i o n s were of the  and  present.  taken  variables  articulatory  contaminants,  using  or  used  inherent  samples,  which would reduce  these e f f e c t s ,  rate  and  the w a l l s was  to  of  phonemes, i s  to minimize  on  the  speech  the r e c o r d i n g s  in i t s  measure  decrease  produced  effects  sensitive  o f samples  individual order  other  is  the  equipment  Furthermore,  measure  both  unrecognized  in  of i n t e r e s t clarity. and  order to  - amplitude  unaccounted  speech  samples.  f o r , may  the  modulation  Nevertheless,  T h i s i s a h a z a r d o f u n d e r t a k i n g an  natural  maximize  other  have b e e n  e x p l o r a t o r y study  6  CHAPTER  TWO  LITERATURE REVIEW  2.1  INTRODUCTION This  speech  chapter  Tests  perception  as  a survey  intelligibility,  subject.  here.  is  o f speech  o f speech  (These  and  tests  Knowledge  intelligibility  and  setting  construction classrooms, aids.  potential many  at  listeners are  the are  most  such  which  the  assessing speech  lecture and  employ t h e s e  progress.  have been  in  a  and  halls, hearing methods  Thus,  intelligibility  s u r p r i s i n g l y , there  degrade  design  systems,  may  useful  influential  to  address  pathologists  of  not reviewed  h a s many  factors  contribute  or  the  quantifying  o f word d i s c r i m i n a t i o n  intelligibility  public  diagnoses  applications  the  measures a r e  many  different  developed.  INTELLIGIBILITY  FACTORS INVOLVING THE SPEAKER  Several  studies  intelligibility the  aimed  on  of  others).  of  FACTORS AFFECTING SPEECH  2.21  literature  of a c o u s t i c a l l y optimal auditoria,  and, n o t  measures  2.2  can  telephones,  making  among  which  Speech-language  when  related  tests  speech  applications.  a v a i l a b l e measures  recognition  include  study of  particular  of  by i m p a i r e d  t h e CID W-22 word l i s t s , The  of the  o f an  literature.  necessary  of  for this  the acoustic  variables  i n d i v i d u a l ' s speech  An e x h a u s t i v e survey,  but  list  of  which  affect  a r e t o be f o u n d i n these  a description  studies  i s not  of p a r t i c u l a r l y  relevant  studies w i l l  approaches  which  the  of  speech  illustrate  have  to  the reader  been e x p l o r e d .  hearing  impaired  accounted  listeners'  judgments o f the c h i l d r e n ' s  between  were:  / t / and  the  second  for  20.5%  absence  /d/  of  variables  and  intelligibility / g / and  vowels, the  and  to  extent  Hudgins,  Ando  and  of the  in  In  voice  / i / and  Canter,  1969)  intelligibility  /b/,  vowel.  first  second  such  or  the  (1983)  regards the  source  (glottis),  (1960)  model o f  and  speech.  and  a l l  between  differences  to /k/ in  previous  variables  affecting  1965;  rate,  and  Numbers,  Brannon,  significantly These  and  1966;  to  findings,  the  however,  (1978).  subject speech  initial  at  duration,  Howarth,  or  cited  Hudgins  contribute  (normal)  f r e q u e n c y change i n  Monsen  as  (2)  other phonetic  not  formant  1938;  time  (accounted  syllable The  formant  addition,  of  were n o t c o n f i r m e d by Monsen  /o/  onset time d i f f e r e n c e s  parameters  John  onset  of the v a r i a n c e ) ;  the authors c l a i m e d that  1960;  Approaching  These  little  frequency (Voelker,  1942;  relative  contribute  /ai/.  prosodic  fundamental  intelligibility.  between a  following  were v o i c e  i n which  acoustic  the  acoustic normal  (3) t h e p r e s e n c e  change  studied  Three  variance  the  in  between and  between /p/ and  diphthong  studies  difference  spectral  found  the  the d i f f e r e n c e s  the v a r i a n c e ) ;  or nasal  of  ( a c c o u n t e d f o r 48.5%  formant  of rapid  liquid  and  (1)  73%  (1978)  children.  variables  differences  for  Monsen  some o f  differently,  signal  the v o c a l  as t h e tract  speech p r o d u c t i o n .  Ananthapadmanabha convolution  filter,  The  after  parameters  of  of  the  Fant's speech  8 originating dynamics", plosive  a t the  source, c o l l e c t i v e l y  c o n s i d e r e d by contrasts,  variations. d y n a m i c s by In  Ananthapadmanabha, a r e  intensity  Formant  changes,  information i s  the v o c a l t r a c t  order to  isolate  the  source  through  a so-called  corresponding signal. passed kHz,  t o each  Each  a  while  peak o f  peak i s  through then  "epoch" f i l t e r .  speech  and  on  passed  The  and pitch  the  source  exclude  a speech  consists  vocal source  pass  and  input to t h i s  the output  t e r m e d an  low  voicing natural  dynamics  epoch.  t h i r d - o c t a v e band-pass  rectified  "source  filter.  i n f o r m a t i o n , Ananthapadmanabha  i s voiced  and  imposed  formant  filter  d e s i g n a t e d as  kind of  of pulses  excitation  The  signal  filter  filtered  signal  i n the  is  first  centred  at 4  w i t h a 340  Hz  cutoff  frequency. Even  without  information result, much  the  f o r the  information When  than  together  be  remained.  a  s o u r c e d y n a m i c s has  a  i n the p e r c e p t i o n of the  phonetic  conclusion i s  difficult  by  the  two  Monsen's  contrasts,  authors most or  (1978) to  w o u l d have  evaluate,  important voice  i n Ananthapadmanabha's  hand, Monsen  results,  overlap.  Monsen w o u l d have p r e d i c t e d t h a t  maintained  the o t h e r  that  phonetic As  Monsen's  included  voicing  differences),  speech  with  groupings  dynamics  (consonant  On  of  enough  previously believed.  taken  variable  would  comprehension  to play  Ananthapadmanabha's  source  information,  Ananthapadmanabha c o n c l u d e d  stronger role  the  formant  as  Since variable  onset  time  intelligibility  processed  incorrectly  speech.  predicted  loss  of  formant  intelligibility  information,  Metz,  and  factors  speakers.  confirmed h i s  labeling speech  affecting  information  concert  with  prosodic  f e a t u r e s a r e an  previous  differences speakers studies  speech,  study,  Picheny  clear  and  sentences of  increased  defined  produced  a l . found  was  in  produced a  clear in  reflection  d u r a t i o n of  as  speech  by  speech  (1)  almost  individual  both  "clear" for  environments.  normal  the  hearing  In t h e i r  1985  between  duration  of  twice the d u r a t i o n speech,  additional  speech  same these  of  differences  conversational of  the  intended  noise.  speech:  instead  "conversational"  intended for  the f o l l o w i n g  with  Picheny,  particular,  with  competing  in  that  dealt  produced  in noisy  speech  "conversational"  sentences  difference  et  have  In  or produced  of  of  dimension.  characteristics  b e i n g speech absence  suggest  investigated  of speech  clear  study  dimension  al.  point  1986)  the a c o u s t i c  contrasted  i n the  et  situations.  was  hearing  intelligibility.  (1985,  speech  the l a t t e r  listeners  this  in  of  also  Monsen, however, and  secondary  to  (1985)  w i t h Monsen i n  primary  important  impaired l i s t e n e r s  authors  the  Metz  different  Clear  hearing  as  in intelligibility  f o c u s e d on  speech.  The  Braida  in  excluded  Monsen's  They a g r e e  authors,  differences  and  Whitehead  replicated  Contrary to  s t u d i e s mentioned  Durlach  and  intelligibility  These a u t h o r s  intelligibility.  inter-speaker  Sitler,  major f i n d i n g s .  segmental  The  Ananthapadmanabha  w h i c h Monsen c o n s i d e r e d i m p o r t a n t .  Sama, S c h i a v e t t i ,  investigated impaired  when  sounds  and pauses  i n the  this and clear  10  condition; and  and  (3)  the short-term  greater  in  results  the  i n d i v i d u a l speech  speech, consonant  relation speech.  of t h i s  preliminary  i n c o n v e r s a t i o n a l speech than i n c l e a r  s p e c t r a of  in clear  conversational  i n s t a n c e s of vowel r e d u c t i o n  there were d i f f e r e n c e s between the c o n d i t i o n s  example,  et  were more  consonant d e l e t i o n  speech; in  (2) there  to  study may  for  i n t e n s i t i e s tended to be  neighboring  As a  sounds;  vowels  c a u t i o n a r y note,  have been  than  in  however,  the  a f f e c t e d by some of  the  i n s t r u c t i o n s g i v e n to speakers b e f o r e r e c o r d i n g of  "clear" condition,  as the  f o l l o w i n g excerpt  from Picheny  a l . (1985) i n d i c a t e s : "The t a l k e r s were a l s o t o l d to enunciate consonants more c a r e f u l l y and with g r e a t e r ( v o c a l ) e f f o r t than in c o n v e r s a t i o n a l speech and to a v o i d s l u r r i n g the words t o g e t h e r . " (p. 97)  These  instruction  probably  number of pauses and Picheny added  f i n d i n g s that  spectrum and  et a l .  l e v e l was  introduced  other a r t i f a c t s  long-term  fundamental  frequencies  speech.  Again,  (1985)  view t h a t  their RMS  not s u b s t a n t i a l l y  c o n v e r s a t i o n a l speech, and in  Picheny et  1985  (root  clear  results, mean  than  square)  a wider range of  in  a l . ' s work supports  conversational Metz et a l . ' s  p l a y an important  of speech.  and  d i f f e r e n t between c l e a r  t h a t there was  prosodic features  the r e l a t i v e i n t e l l i g i b i l i t y  disproportionate  i n t o the speech samples.  (1986) confirmed the  a  part i n  11 2.22  FACTORS INVOLVING THE French  which of  and  Steinberg  can a f f e c t  the s i g n a l ,  and  phase  background  distortion. filtering  filtering,  in its  the  effect  of  of  the high  of  the m a j o r i t y  pass  prevalence transmission,  were  to  be  the None  the  an  These place  of the lower  were  at very  the purposes  was  this  of speech  Voicing  under  particularly case  to  the  speech is  dimensions  affrication, and  nasality  subjected  to  of a r t i c u l a t i o n  was  these  conditions.  resistant  to high-pass  the remaining and c o n s e q u e n t l y  perception.  and  filtering  most o f t h e a c o u s t i c  intensity,  low-  laboratories.  nasality,  whereas p l a c e  dimension  the  the  i n d e t e r m i n i n g speech  s p e e c h was  removed, l e a v i n g low  barrier  "articulatory"  r o b u s t when  disrupted  was  five  to  intensity  ( i . e .with  high-pass  voicing,  filtering,  in  than  and  low-pass  audio systems, a  of a r t i c u l a t i o n .  o f the dimensions  consonants  low-  Given the nature  outside acoustic  defined  be most  since  as  important f a c t o r  Nicely  most e a s i l y  information for  and  or low-pass  filtering,  reverberation,  i s roughly equivalent  fidelity  rather  factors  intensity  out t h a t  hearing losses  noise  in situations  found to  noise  speech,  o f low  low-pass,  and  the  (1955) added  but p o i n t e d  n o i s e because  background  speech.  duration,  Nicely  of  o f damage t o h i g h f r e q u e n c y h e a r i n g ) , effect  intelligibility  in  on  of sensorineural  of  Miller  include  f r e q u e n c y components o f s p e e c h .  degree  likely  These  and  to the l i s t , effect  a number  n o i s e i n the system,  Miller  background  filtering  more  SYSTEM  (1947) l i s t e d  intelligibility.  high-pass  greatest  TRANSMISSION  energy i n available inaudible  12  Intelligibility reverberation.  o f speech  i s also  Reverberation  adversely  i s defined  a f f e c t e d by  by R e t t i n g e r  (1968)  as: "sound persistence due to repeated boundary reflections a f t e r t h e s o u r c e o f s o u n d has s t o p p e d . " ( p . 85) Boundaries  in  ceilings,  or  this any  intelligibility sound  energy  case  are  object  surfaces  in  an  i n overlap  as  enclosed  i s r e d u c e d by r e v e r b e r a t i o n results  such  walls  space.  or  Speech  because p e r s i s t i n g  of successive  speech  sounds  and b l u r r i n g o f t h e s i g n a l . The  determining f a c t o r i n the s u s c e p t i b i l i t y  degradation identified define  through  reverberation  as the r e v e r b e r a t i o n  reverberation  has  time.  been  o f speech t o traditionally  Morse a n d I n g a r d  (1968)  time a s :  "The l e n g t h o f t i m e i t t a k e s t h e mean e n e r g y o f t h e wave t o reduce to a m i l l i o n t h part of i t s i n i t i a l mean v a l u e " , ( p . 5 5 8 ) , or,  i n other  decrease that  words,  b y 60  speech  reverberation  is  and  0.8 s e c o n d s . changes  a  and Burger  by  partially  Morse  reverberation  time  f o r normal  but that  only  at  t h e s i g n a l and  add t h a t  i f the  0.3  speech  i n a t i m e l e s s t h a n one t e n t h o f signal will  Furthermore, of  to  (1964) c o n c l u d e d  i n t e g r a t e d a t t i m e s between  time, the o r i g i n a l  energy.  wave e n e r g y  reverberation  and I n g a r d  significantly  sound  intelligibility  unaffected  are  reverberation  reflected  Lochner  f o r the  t i m e s b e l o w 0.3 s e c o n d s ,  reflections  the  time taken  decibels.  its  signal  the  1.2  Crum (1974) s t a t e d  seconds  hearing  be b l u r r e d by  adults  or  more  i n quiet,  that  decreased and t h a t  13 the  combination  speech sum  background noise  recognition  of  2.3  of  the  performance  e f f e c t s of  the  and  reverberation  more t h a n  predicted  reduced  from  the  individual variables.  SPEECH I N T E L L I G I B I L I T Y MEASURES BASED ON  LISTENER  JUDGMENTS 2.31  INTELLIGIBILITY Perhaps  are  the  most  intelligibility  (1987),  w h i c h may  SCALES subjective  scales be  measures  s u c h as  used,  for  the  two  example,  population  of  speakers  for i n t e l l i g i b i l i t y  and  3  describe  these  to  Figure the  quick  factors  to  score  Doyle  studied  assessing  hearing  indicated  good  reliability, speech concern  the  use  impaired  particularly  i n the  use  e m p l o y e d by for  Doyle  screening  deficits. no  a  Figure  2  d e t a i l s as  intelligibility,  but  are  administer.  intra-rater  of c e r t a i n  intelligibility  s c a l e s , which g i v e  affecting relative and  of  of  scales  children's  for  the  but  scores  Thus,  intelligibility  by  speech.  reliability,  individuals. of  these  The  results  poor i n t e r - r a t e r assigned  listener  scales.  audiologists  bias  to  the is  a  Speech l a completely unintelligible.  Figure  2.  Speech I s very d i f f i c u l t to understand, only Isolated words o r phrases are intelligible.  The  National  With d i f f i c u l t y Che l i s t e n e r can u n d e r s t a n d about h a l f o f t h e message (Intelligibility may improve after a listening period).  Technical  Intelligibility  (after  Institute Doyle,  Speech I s Intelligible w i t h the exception of a few words and p h r a s e s ,  f o r the  1987).  Deaf  Speech I s completely Intelligible,  (NTID) S c a l e  of  111111111111111111111 0  1°  20  30  40  50  60  70  80  COMPLETELY UNINTELLIGIBLE  F i g u r e 3.  90  100  COMPLETELY INTELLIGIBLE  A percent r a t i n g s c a l e o f i n t e l l i g i b i l i t y  ( D o y l e , 1987).  16  2.32  DETAILED JUDGMENT-BASED INTELLIGIBILITY TESTS  Black Appendix of  (1957) reviewed A for  a multiple  a list  early i n t e l l i g i b i l i t y  tests  (see  of the t e s t s ) and he advocated the use  c h o i c e format  with answer forms p r o v i d e d .  For  example,  a listener  would be  presented with a c h o i c e of four  possible  words, and  he would  circle  heard.  He  listeners  on  wrote  down  words  W i l l i a m s and aimed a t  Hecker  The  of r e d u c i n g the  burden  the assessment A).  speech  distortion  (additive  and v o c o d i n g ) ,  conditions. a l . (1954)  varied  of speech  speech-shaped  and two  individual  introduced  Furthermore,  t r a n s m i s s i o n systems  These authors used v a r i o u s types of  the s c o r e s  into  noise,  d i f f e r e n t speakers  the  same  obtained f o r  peak f o r the  f i n d i n g of H i r s h  intelligibility  r e l a t i v e to one another depending  distortion  (nor n e c e s s a r i l y more  They confirmed the e a r l i e r  that  time,  (1968) compared the r e s u l t s of f o u r  i n Appendix  test  they  tests.  (listed  clipping,  he  heard.  otherwise they were no more r e l i a b l e  tests  thought  knowledge of the s c o r e r and the s c o r i n g  v a l i d ) than the write-down  et  the  c h o i c e t e s t s had the advantage  the p h o n e t i c  but  word he  compared the m u l t i p l e c h o i c e format to t e s t s where simply  multiple  the  test  scores  on the type of speech transmission  t h e i r two  system.  speakers were  more s i m i l a r f o r some d i s t o r t i o n types than f o r o t h e r s . More chapter tests listed  r e c e n t l y , Newman c o n s i s t i n g of  used by  (1979) wrote the i n t r o d u c t i o n to a  a c o l l e c t i o n of reviews of a r t i c u l a t i o n  speech-language  i n Appendix  A. These  pathologists. t e s t s were  The  tests  are  developed to r e p l a c e  17  spontaneous and  speech samples with the i d e a of d e c r e a s i n g  analysis  times,  and  ensuring  sampled  in a  scales  and  (1957),  articulation tests,  performed  given s e s s i o n . the  In  intelligibility  that  a l l phonemes  phoneme  intelligibility.  c o n t r a s t to i n t e l l i g i b i l i t y tests  as w e l l  discussed  confusions Still,  which  these t e s t s provide  only  phonetician  cannot analyze  summarize,  Newman  questionable  phoneme e r r o r s  criticized  validity  acknowledged  and  recent e f f o r t s  these  to improve  superficial  acoustically. tests  for  To their  although  he  this situation.  In  he applauded the a d d i t i o n o f suprasegmental phonemes  the content  o f some  t e s t s , whioh  validity.  2.4  ACOUSTIC INTELLIGIBILITY MEASURES  should  help  te i m p r o v e  THE ARTICULATION INDEX (AI)  The Steinberg  Articulation (1947).  telephone  purposes,  Index  (AI) was  conceived  Although o r i g i n a l l y intended  systems,  the AI  has been  revised  The summarized  by French and f o r assessment for several  and i t i s w e l l enough e s t a b l i s h e d to be d e s c r i b e d i n  an American N a t i o n a l Standards I n s t i t u t e standard  of  affect  a w e l l .trained  reliability,  test  of  Black  analyses  adversely  i n f o r m a t i o n , s i n c e even the ear of  2.41  by  as p h o n o l o g i c a l  descriptive  to  were  on spontaneous speech samples, g i v e a t t e n t i o n to the  specific  addition,  testing  Articulation  Index  was  speech i n t e l l i g i b i l i t y  l i s t e n e r judgments,  an  innovation  (ANSI 1969). in  that  it  i n t o one number, independent  and was f i r m l y based on a c o u s t i c s .  As  18 originally  f o r m u l a t e d by  Articulation  Index i s  input  spectrum  speech  output of  o f a speech  n o i s e and  divided each  w i t h good  where  sharp  where  there i s  factors,  Pavlovic modifications well  resolution, speech or  authors  Ahlstrom,  (1987)  effects is  200-6100  Hz,  signal  speech  bands p r o v i d e s a in  situations  takes place,  specific  or  background  ( e g . K r y t e r , 1962a;  ANSI  and  have  Kincaid,  1986)  weighting  obtained.  provided  a  summary  w h i c h have b e e n a p p l i e d  Index  actual  spectrum  bands, t o g e t h e r w i t h  scores  tables  the  ideal  e q u a l l y to  useful  frequency  bands o r o c t a v e  as updated  Articulation  Bell,  AI  speech  to c o n t r i b u t e  However, l a t e r  similar  The  the  an  c o n s i d e r i n g the  bands i n t h e r a n g e  of the  n a r r o w band  octave with  determined) with  of twenty frequency  frequency  Humes, D i r k s , 1/3  use  filtering  interference.  used  frequency  The  (1947),  o b t a i n e d by c o m p a r i n g  (empirically  i s considered  intelligibility.  Steinberg  t r a n s m i s s i o n system  twenty  of which  1969;  a ratio  and  band-pass f i l t e r i n g .  into  measure  French  of v a r i a b l e s  is calculated  by  of  t o t h e AI f o r AI  the  various  s i n c e 1947,  calculations.  means o f t h e  following  as The two  equations: A  =  PlljW,-  s = T (A ) , where  A  is  the  Articulation  function  (a measure t a k i n g i n t o  speaker  and  materials), the  the i is  proportion of  familiarity the  frequency  the speech  Index,  P  account of  is  the  proficiency  the e n u n c i a t i o n of  the  listener  with  the the  band u n d e r c o n s i d e r a t i o n , W- i s dynamic range  within  frequency  19  band  i  which c o n t r i b u t e s  over  the  transmission  that  frequency  to  the  AI  second  the  original  remedied  this  room  (1986)  that  found  to-noise  ratios  stressed  that  single  intelligibility  related in  the  provision  (1962a)  partly  p r o v i d i n g c o r r e c t i o n f a c t o r s based  which,  in  modified  ANSI s t a n d a r d .  these  form,  However,  than  zero  d e c i b e l s , and  c o r r e c t i o n s were  b a s e d on  were  Humes e t a l .  c o r r e c t i o n s were i n a d e q u a t e a t  worse  the  Kryter  no  signal-  Kryter  the  himself  results  of  a  study.  The  validity  investigations. scores  of  speech  compared  pointed  materials  intelligibility material  subjects  to  AI and  to  has  been t h e  Steinberg  listener  speech m a t e r i a l s out,  greater  results score.  i n which  i n which  the  French  a v a r i e t y of  (1962a,b)  or  by  time  i n t o the  using  c o n t r i b u t i o n of  Index c o n t a i n e d  conditions.  situation  reverberation  AI  ideal  empirical t r a n s f e r function given  incorporated  of  intelligibility-  i s the  s i s speech  Articulation  reverberant  on  and  I'  speech  equation.  The for  system,  band,  through  to o v e r a l l  a few guess  in a By  words a r e at  (see  words  be  more  4).  a  score  As  Kryter of  f o r any  redundancy,  chart  obtained  redundancy  AI  gleaned  repeated  numerous  scores  Figure  semantic  semantic  of  (1947) p r o v i d e d  judgment  smaller  meaning c a n  subject  he  the given meant  from s y n t a c t i c cues, often,  which  accurately.  allows This  0  0.1  0.2  0.3  0.4  0.5  0.6  0.7  0.8  0.9  1.0  ARTICULATION INDEX  Figure 4. R e l a t i o n between A r t i c u l a t i o n Index scores and PB-word scores, (reproduced from French and Steinberg, 1947).  means any  t h a t an  AI score  has l i m i t e d meaning i n i s o l a t i o n  from  s p e c i f i c a t i o n o f the speech m a t e r i a l to which i t a p p l i e s . Pavlovic  Bell,  Rossman  validity They of  (1984), Kamm, and  D i r k s and B e l l  Kincaid  (1986)  (1985), and D i r k s ,  have  o f the AI when a p p l i e d t o hearing  investigated impaired  the  listeners.  found t h a t the AI was a good p r e d i c t o r of the performance most l i s t e n e r s ,  severe high  with some  exceptions  among  frequency s l o p i n g s e n s o r i n e u r a l  subjects  hearing  with  losses.  2.42 THE SPEECH TRANSMISSION INDEX (STI, RASTI) Houtgast index  and  having one  Steeneken  b i g advantage  provisions  f o r taking  filtering,  and r e v e r b e r a t i o n ,  were  in,  built  factors.  introduced  over the  i n t o account  r a t h e r than  Instead  artificial  (1971)  on  acoustic  A r t i c u l a t i o n Index:  peak c l i p p i n g ,  as w e l l added  an  band-pass  as  background  as  clumsy  noise,  correction  o f n a t u r a l speech s i g n a l s , the STI employed  signals  - another d i f f e r e n c e from the A r t i c u l a t i o n  Index. The a of  o r i g i n a l Speech  weighted sum  o f spectrum d i f f e r e n c e s between the two l e v e l s  an a l t e r n a t i n g  signal in  centred  at frequencies  signal  level difference  system  was compared  principle, Figure  5.  T r a n s m i s s i o n Index (STI) c a l c u l a t e s  as a p p l i e d The two  each o f  ranging  from  the f i v e 250 to  4000  a t the input t o a speech  to the  d i f f e r e n c e at  to background levels  of  the  noise,  octave Hz.  bands The  transmission  the output.  This  is illustrated in  alternating  signal  are  22 referred  t o as  two  separate  signals  in  the d i s c u s s i o n  which  follows. At to  the input,  resemble  signals but  input,  1)  the  was  signals  the i n t e n s i t y  equal  across  were  passed  background their  the  had  significantly essentially  a  level  octave  bands,  noise,  since  3 and  would  i t . 1  these d i f f e r e n c e s  different  contribution  at input)  be  would  intensity Sound  differences  sum  (3 Hz  across  be  result  and a  of  noise, the  i n an  spectral  1, however,  of  between  the would  o f each o c t a v e  would  containing  ( A L ) a t the o u t p u t would  effects  weighted  rate  signals  being  would  be  spectrum  a t the o u t p u t , i n each o f the  the  present  was  2,  Comparison  be e q u a l  was  the  system  than the background  reveal  the  1  longer  reverberation  however,  with the n o i s e ,  Sound  at  t o Sound  o f Sound 2.  by  the i n t e n s i t y  of  If,  3) h a v i n g h i g h e r  from t h a t  of the  (Sound 2 ) ,  shape  equal i n i n t e n s i t y  difference  unaffected  o f Sound  spectral  shaped  One  than the other  transmission  more i n t e n s e  levels  If  of noise  between them a t t h e i n p u t  frequency range.  (Sound  different  no  consisted  t h e same  Sound 2, c o m b i n e d signal  would  intense  noise approximately  changed.  shape  more  difference  through  intensity  output  signals  the average long-term speech spectrum.  (Sound  since  t h e two  the  background  t h e two  frequency  signals  range.  take i n t o  A  account the  band.  i n a system, changed  after  the  alternation  transmission.  INPUT  >  SPEECH TRANSMISSION  SOUND 1 and SOUND 2 alternating  SYSTEM  •>  OUTPUT SOUND 1 and SOUND 3 alternating  BACKGROUND NOISE  sum o f SOUND 2 & BACKGROUND NOISE  1—>  •H 0.25  A  Frequency  (kHz)  F i g u r e 5.  >  1 0.25 Frequency (kHz)  h  4  _^ 0.25  1  >  A  Frequency (kHz)  The way i n which the o r i g i n a l Speech T r a n s m i s s i o n Index accounted f o r the e f f e c t s of background n o i s e .  24 Reverberation alternation The were  was  rate  from  signals,  developed  measured input to  and  and  transmission  channels.  The  were u s e d  formula  i is  level  differences, difference the  The Index  (MTF)  Houtgast, in  by  originated the  the  The  effects  for  STI, seven  measured and  over  to fifty  combinations  i n the  early  of  noise,  and  channels.  form  was:  &  of  concept,  level  indicated  for  frequency  intensity.  the  Modulation 1973;  whereby  Transfer  Steeneken the  background  output noise,  p e r c e p t i o n (see F i g u r e 1).  frequency  of bands  Modulation of  a  and  fluctuations  a r e smoothed i n t h e  r e v e r b e r a t i o n and  influence  by  o f the Speech T r a n s m i s s i o n  Steeneken,  of v i s u a l the  account  or extremes of  rate  intensity  Limitations  to  intensity  alternation  initial  signals.  input signal of  i s the output  i s the  inability  MTF  o f an  i n studies  revised  Functions  two  (Houtgast  the envelope  signal  20 dB  incorporation  1980).  which f o l l o w e d ,  interfering  index,  s t e p i n the e v o l u t i o n  the  in  of comparisons  incorporates  center clipping,  was  Function  (which  included  next  in this  band  between t h e  distortion,  change  1^)  ^<  i f any),  authors  filtering,  as contaminants  the octave  difference  the b a s i s  word s c o r e s  f o r t h e STI  = 1/5  procedure  Various degrees  band-pass  STI  where  (PB)  the  output.  m o d i f i e d on  Balanced  peak c l i p p i n g  quantifying  the a n a l y s i s  Phonetically  reverberation,  by  In  Transfer  sinusoidally  25  modulated a  single  input signal score.  Articulation with in  of shaped pink  The  calculation  Index.  The  during times  correlation  measurements, will  result  and  and  is  i n Houtgast  presented The  applications  authors  have  Speech T r a n s m i s s i o n The Steeneken field  first  Steeneken  meter,  in  (centered  is  calculation  Steeneken  noise  reverberation review  of  in  the Index  (1985).  three  the development  directions  with  the  1985,  two  and  Dareham,  of  1986)  &  meter used  Steeneken,  RASTI  octave  are  similar  STI  This  the  2000 Hz)  i s otherwise  an  (Houtgast  introduced.  only Hz  o f measurement d e v i c e s .  (1978) d e s c r i b e d  I n d e x ) was  a t 500  illustrated  o f the Speech T r a n s m i s s i o n  More r e c e n t l y ,  which  A  the  scores  background  dependent  scores.  from  STI  is  into  Index.  & Houtgast,  Transmission  STI  moved  & Agterhuis  studies.  and  of r e v i s e d  changing  frequency  in invalid  adapted  word s c o r e s  Non-linear d i s t o r t i o n ,  development  of  scheme was  (Dutch) P h o n e t i c a l l y Balanced F i g u r e 6.  n o i s e a r e combined  (Rapid  is  a  1984; Speech  screening  frequency  i n c l u d e d , but  to the Speech  in  the  bands method  Transmission  Index. The optimal point  direction  auditoria, (Houtgast,  Houtgast for  second  and  using  of  the  the  a desired  Steeneken  Steeneken  t h e volume  is  (1980)). room,  and The the  design STI Plomp  value  of  acoustically as  (1980)  a  starting  and  Plomp,  specifications  given  were  reverberation  time,  the  26  100 k  a  V  80  *  * 60  . 1 1  • 1  o 3 i  •  1  1 1  A  m o. 20  f 20  1 . =N.BP . = p e a k c: l . • = A G C / FI E V a = 5.6% LQ  60  80  100  STM%)  F i g u r e 6.  R e l a t i o n b e t w e e n S T I a n d PB-word  score  (Dutch  words)  f o r the  c o n d i t i o n s w i t h n o i s e , bandpass l i m i t i n g , peak c l i p p i n g , automatic g a i n c o n t r o l , and r e v e r b e r a t i o n . The c u r v e r e p r e s e n t s t h e b e s t f i t t i n g curve f o r a l l these data p o i n t s . (Reproduced from S t e e n e k e n and H o u t g a s t , 1980).  27  ambient  noise l e v e l ,  the o r i g i n a l  signal intensity,  and the  d i s t a n c e between speaker and l i s t e n e r . The tracing  t h i r d new  model designed  individual the  d i r e c t i o n f o r the STI to  provide  audience p o s i t i o n  an  i s a computer r a y -  STI  score  for  each  r a t h e r than merely one s c o r e f o r  e n t i r e a u d i t o r i u m (van R e i t s c h o t e , Houtgast and Steeneken,  1981,  1983).  In t h i s  model, the  speaker i s  p o i n t source which emits a s i g n a l to each 2.43 MODIFIED SPEECH TRANSMISSION The  mSTI i s  combines  the  Articulation frequency an  transfer  Humes et i t s two  intelligibility  the STI i n that i t  function  approach  The  the STI.  predecessors  speech was  mSTI d i d  by  for prediction nermal  and  The mSTI s c o r e s matched best with s c o r e s o b t a i n e d  (1971) and  Articulation  called  because  intelligibility information, consistent  heaping  indeed prove s u p e r i o r to the AI and  2.44 ARTICULATION LOSS OF CONSONANTS  the  of  t e m p o r a l l y and s p e c t r a l l y  by h e a r i n g impaired l i s t e n e r s on a speech r e c o g n i t i o n  Peutz  with  a l . (1986) c r e a t e d the mSTI as  parfermane§  l i s t e n e r s when  distorted. to  AI and  Index weighting f a c t o r s f o r one t h i r d octave band  analysis.  impaired  audience p o s i t i o n ,  INDEX (mSTI)  o f the  modulation  improvement on  speech  a hybrid  s i m u l a t e d by a  Klein  Loss  of  Peutz over  a  and because  (ALcons)  (1971) developed a measure c a l l e d Consonants regarded  transmission the measure  f o r transmission  test.  of  measure the system  (ALcons),  degradation as  proved t o  consonants  than  a  loss  so of of  be much more of  vowels.  28  Like  the STI,  effects, input  ALcons  but, l i k e  signal.  directly  the AI,  accounted  i t employed  Peutz suggested  ( d ^ ) , speech  intelligibility  on  speaker  listener  intelligibility the are  to  i s independent  reverberation used  to obtain  time  t h e ALcons  distance.  room.  = 200 dJ_T*+ a ( % ) , V  for  d 2 dj,  ALcons  = 9T + a ( % ) ,  Above  d  and v a r i e s  c  ,  with  = (0.2 s*m»)J V/T *  these equations,  is  the distance  d  i s the c r i t i c a l  to the  listener  volume  i n m , T i s reverberation  ALcons  i s the i n t e l l i g i b i l i t y skills  recognition  of test.  i s competing Peutz  (1982),  established,  the  this  in  score,  listener,  background  noise  validating even  the  though  V  i n meters, is  the  d  room  ( a t 1400 Hz) i n s e c o n d s , and a i s a c o r r e c t i o n f o r  as  A modification  distance  meters,  time  (1971) u s e d v e r y s m a l l  p e o p l e ) when  Lundin  dependent  v  c  In  there  critical  measure:  ALcons  the  as an  The f o l l o w i n g e q u a t i o n s  d < d , c  speech  is partially  for  with d  ten  natural  of distance,  of the  reverberation  t h a t up t o a c e r t a i n  distance the  for  measured  i s made  by  a  speech  t o t h e measure i f  i n t h e room. groups ALcons.  its  of listeners Still,  validity  measure i s w i d e l y a c c e p t e d .  (five to  according  is  not  to  well  29  2.45 DIRECT-TO-REVERBERANT The as  Direct-to-Reverberant  SRR,  for  combines  Signal  features  similar  to the  incorporate  The  to  Intensity  Reverberation  o f the ALcons with  Articulation  the b u i l t - i n  simplicity AI.  INTENSITY METHOD (SRR)  SRR may  Ratio  It  consideration  was  of  known 1986),  approach  formulated  reverberation  the frequency  be c a l c u l a t e d  (Lundin,  a f r e q u e n c y band  Index.  o f t h e ALcons with  method, b e t t e r  specificity  according  to  the  to and  of the  following  formulae: SRR = - 2 0 . l o g d / r r  where  listener, the  X  the distance r  distance  point are  r  between  original  intense  source  which a c t u a l l y  absorption the  room  %  i n meters  and t h e  reaches  room i n  the  Q i s the d i r e c t i v i t y  of the  o f sound  listeners the  sabins',  and T t h e i s r e v e r b e r a t i o n  s e c o n d s . The two f o r m u l a e  s i g n a l and t h e signal  around  metric  ( d e f i n e d as  reverberant  to the proportion  i s dissipated  o f the in m,  signal  (SRR = 0 d B ) ) ,  (which c o r r e s p o n d s  sound energy  radius  the source o f the o r i g i n a l  source  the  i n m e t e r s between t h e s o u r c e and t h e  i s the r e v e r b e r a t i o n  where t h e equally  0.057 ^ J Q V Z T T  = /QA/16Tr =  d is  r  from t h e  when t h e r e s t o f room), A  i s the  V i s t h e volume o f t i m e o f t h e room i n  f o r c a l c u l a t i o n of r are related v i a r*  The u n i t o f a b s o r p t i o n , t h e s a b i n , i s named i n h o n o r o f W.C. S a b i n e , a n d h a s t h e d i m e n s i o n s o f one s q u a r e foot. A metric s a b i n has the dimensions o f one s q u a r e m e t e r , and i s therefore e q u a l t o 10.76 s a b i n s , s i n c e t h e r e a r e 10.76 s q u a r e f e e t i n one s q u a r e m e t e r .  30  Sabine's  formula: A  where  = 0.163 (s/m) V A T  the absorption  coefficient,  oi , i s assumed t o be e q u a l  t o one. Lundin provide the  (1986) f o u n d ,  better  AI, or  roughly  the STI.  listener  failed  to  than the ALcons,  a l l four  measures  performance o f  conditions,  but  intelligibility  SPEECH COMMUNICATION SCI ( K r y t e r  adopted  as  gave  normal  that  these  measured  by  instrumentation. basis  of  INDEX ( S C I )  and B a l l ,  because  bands,  of the  adverse  overestimated  that  SRR  judgments.  The  the  He c o n c l u d e d  in  the  of i n t e l l i g i b i l i t y  predictions  listeners  predictions  2.46  predictions  equivalent  hearing  however, t h a t  i t s An  the  1964)  use  frequency s h i f t ,  not  requires  intelligibility  signal-to-noise  has  score  ratio  been  widely  sophisticated i s calculated  in  nine  on  frequency  a n d peak c l i p p i n g i n t h e s y s t e m  under  study. 2.47  PATTERN CORRESPONDANCE The  another analysis this speech  PCI ( L i c k l i d e r , index which because o f  case,  transmission  Bisberg  has o n l y  This  method  applications  instrumentation  of the running  compared b e f o r e  system.  and S c h w a r t z l a n d e r ,  specific  t h e complex  the pattern  signal i s  INDEX ( P C I ) 1959) i s  i n systems  required.  In  power s p e c t r u m  of a real  and a f t e r passage  through a  inspired  Houtgast  and  31  Steeneken  (1971)  Transmission  2.48  recent  A  still  in  microphone numerals  ratio,  articulation  the  Speech  i t s infancy,  is  by  average  with disorders to  a  this  amplitude, time,  such as  microcomputer  the p a t i e n t .  inter-syllable  measured as  He i n t e n d s  t o be u s e d by s p e e c h  connected  spoken  o f speech a t  The measure  peak t o a v e r a g e and  t h e number  of s y l l a b l e s  a  number  speed  of  produced i n  seconds.  2.49  MONSEN'S FORMULA Having  identified  influential  variables  intelligibility,  Monsen  in  small the  (1978)  2.21)  developed the  speech  o f the h e a r i n g impaired:  I  = 0.91(T  t  d  I is  onset  time o f  t h e mean  following  - T ) + 0.0214(F  where  is  measure.  of patients  peak a m p l i t u d e ,  amplitude  ten  is  digitizes  considers  created  intelligibility  (1985)  f o r assessment  dysarthria. which  to quantify  Kondraske's  which  clinicians  originally  METHOD  attempt  source i s  method,  they  Index.  KONDRASKE'S  A its  when  ;  second  (discussed formula  - F) 0  above  t o be  of  speech  in  section  applied  to  the  + 4.78(L,N) + 54.57,  i s t h e mean v o i c e  formant  especially  determination  the index o f i n t e l l i g i b i l i t y , / t / , T(j  of  T  t  i s t h e mean  voice  onset time of /d/,  frequency f o r / i / ,  F  3  F,*  i s t h e mean  second  formant frequency  variables change  i s an  by  initial  e m p i r i c a l l y determined  v a l i d i t y of  between  /,  L  and  r e f l e c t i n g the presence or absence  following syllable  54.57 the  f o r /J  h i s formula,  liquids  N  numerical  of r a p i d and  constant.  spectral  nasals,  and  Monsen t e s t e d  and found a c o r r e l a t i o n of  p r e d i c t e d and o b t a i n e d i n t e l l i g i b i l i t y  normal h e a r i n g l i s t e n e r s .  adopted  are  0.86  scores assigned  T h i s formula has not been w i d e l y  elsewhere, however,  probably i n  p a r t due to the time  consuming s p e c t r o g r a p h i c measurements r e q u i r e d .  2.5 COMPARISONS OF ACOUSTIC INDICES Some limited  of the  indices described  a p p l i c a t i o n s and  their  introduction.  which  has the  language to  popularity  o f f i c e microcomputers,  since method, speechhas yet  But what of the mSTI, the STI, the AI, the  and the SRR? of these  linear  frequency or  signal  intensity.  SRR  gained  being a v a i l a b l e to c l i n i c a l  p a t h o l o g i s t s through  None  to  not  The a p p l i c a b i l i t y of Kondraske's  promise of  be determined.  ALcons,  have  (Monsen's, PCI, SCI) have  the AI  equipped to  The STI, mSTI, ALcons,  an  deal  with  non-  amplitude d i s t o r t i o n , or with extremes  for reverberant  require  interfering  measures i s  and SRR  of  are s u p e r i o r  c o n d i t i o n s , but the ALcons and the  external  correction  in  the  presence  of  noise.  Humes  et a l . (1986) found the mSTI to be s u p e r i o r to both  the  AI or  the STI  the  presence o f  i n p r e d i c t i o n of i n t e l l i g i b i l i t y  temporal and  spectral distortion.  scores i n However,  33  they  a l s o found  underestimate subjects, et  a finding  through  which i s  and P a v l o v i c  t h a t the  predicted  three of  these measures  loss of i n t e l l i g i b i l i t y  a l . (1986)  found  that a l l  AI, the  i n some hearing  impaired  i n agreement with those of Kamm  (1984).  STI, the  S i m i l a r l y , Lundin  SRR,  higher i n t e l l i g i b i l i t y  listener  tended t o  and  scores than  the  (1986)  ALcons a l l  those  obtained  judgments.  2.6 FURTHER DEVELOPMENTS OF THE MODULATION TRANSFER FUNCTION The of  Modulation T r a n s f e r F u n c t i o n has aroused  others b e s i d e s  In  1981,  for  Modulation Fourier  Steeneken, Houtgast instance,  the i n t e r e s t  and t h e i r  Schroeder  described  colleagues. the  Complex  T r a n s f e r F u n c t i o n (CMTF), which i n v o l v e s the use o f transforms,  differences  and  includes  consideration  together with r e d u c t i o n i n modulation  of  depth  phase in its  calculation. Elsewhere, 1983,  1985; Ahlstrom,  method  or v a l l e y s  They  have  aids .  Humes, 1985)  psychoacoustic  t h r e s h o l d s f o r temporal  peaks  subjects  and h i s c o l l e a g u e s (Ahlstrom & Humes,  Boney &  f o r assessing  behavioural  and  Ahlstrom  of  sinusoidally  investigated  have developed a  MTFs  by  probe tones modulated  Modulation  Transfer  obtaining  (tone p i p s a t  speech  Functions  with normal h e a r i n g and s e n s o r i n e u r a l h e a r i n g  subjects using  compression  and  noise).  non-compression  in  losses, hearing  3  2.7 CONCLUSION Even  given  a l l of  these  Transfer  F u n c t i o n seems  measures  on which i n d i c e s have been based.  both  reverberation  external many  the  and  c o r r e c t i o n s , and  authors working  promising  to  results.  Modulation  be  limitations, the  most  background i t has  the  versatile  o f the  I t can account f o r  noise  warranted  Modulation  effects the  without  a t t e n t i o n of  i n s e v e r a l d i f f e r e n t d i r e c t i o n s , a l l with Perhaps  Transfer  in  a form  Function  y e t to be determined, may  i n t e l l i g i b i l i t y measurement t o o l o f the f u t u r e .  well  be  the  1  CHAPTER THREE METHODS AND  3.1  OVERVIEW OF The  THE  objective  possibility  of  EXPERIMENTAL DESIGN of  this  devising  intelligibility  when  clarity  speaker.  of the  (henceforth evaluated of  referred by a  For  the  articulatory  i t  was  an  measure  to  as  clarity At  to  of  was  end or  there  At  was  the  "Overarticulated"  intelligible  Speakers  producing  three conditions  the  intended level of the  4.12.  speech  measure  MI),  was  judgments  MI  of  in noisy  three  the  range  of  "articulatory there  condition,  "Normally  was  the  which  was  In t h e m i d d l e  Articulated"  range,  there  (N)  was  the  This  condition  corresponded  such  as t h a t  intended for  conditions.  sentence-length to produce  mentioned.  were of  I n most c a s e s , b u t n o t a l l , clarity  speakers  be  will  utterances  the s e n t e n c e s i n each  of a r t i c u l a t o r y  values  the  intelligibility.  speech,  T h e y were a s k e d  the  Section  or  perceptual  the range,  (0) c o n d i t i o n .  listeners  performance  of  the  top  of hearing  recorded.  of  the  articulatory  Index,  into  mumbled  range,  hard  explore  computed  experiment,  divided  t h e low (U)  this  of  maximally  the  of t h i s  listeners'  to c o r r e s p o n d to poor  to  merit  on  Modulation  intended  condition.  only  to  materials.  "Underarticulated"  the  acoustic  The  purposes  conditions".  investigation  depends  comparison  t h e same s p e e c h  MATERIALS  calculated  was  attained.  discussed for  in detail  the speech  The in  samples  36 were in  compared w i t h  the  3.2  form  of l i s t e n e r  PREPARATION OF  3.21  the p e r c e p t u a l  THE  judgments o f a r t i c u l a t o r y  composed. each  Each  sentence there  predominantly  labial  predominantly  alveolar  containing  sentences  3.22  as  many  nine  contained  French  nine s y l l a b l e s ,  three  sentences  consonants, and  palatal  velar  and  English  and  for  containing  consonants, An  were  containing  three  consonants.  French  sentences  and  effort  three  was  phonemes  made  in  the  possible.  SPEAKERS  Ten  English  recorded.  speakers  The  speakers  French  (four  n a t i v e Western  Western British speaker  and  Among t h e E n g l i s h  female.  English  Canadian  dialect.  (Received had  a  one  speakers,  speaker  m a l e s and  Canadian  was four  residents One  unlike  the other  of the speakers  were j u d g e d  he  speaker  Eight  females) who  were  spoke  standard  one  had  speaker  t o have n o r m a l  French  a  female  spoke E n g l i s h ,  recorded  the  long-time  the  and  five  of  speaker  French  also  were  were male and  male.  The  S w i t z e r l a n d ; he  ten speakers,  five  accent,  Newfoundland a c c e n t .  o f Lausanne,  French  male E n g l i s h  Pronunciation)  native  All  and  were  predominantly  r e p r e s e n t as  or  clarity.  SPEECH SAMPLES  nine English  language  were  quantified  SPEECH MATERIALS  Initially,  to  d a t a , w h i c h was  was  a  but,  sentences.  speech.  3.23  RECORDING OF SPEECH SAMPLES Speech  acoustic AKG  s a m p l e s were  tiling,  using  D202 d y n a m i c  sentences  each  Underarticulated  (U  the speaker  Overarticulated  to get  in  Underarticulated  condition.  condition  were  not  Listening  test.  last  recording speaker  the other  (Normal).  condition  was i n t e n d e d  patterns  of  might  felt  produce  that  Overarticulated  used  first  to  a f t e r the  recording  the second  the  Normal  or  i n the  was  recorded  i t w o u l d be t h e most  intentionally,  and  would p o s s i b l y  that  help the  This  to  recording  was  done  because  to r e f l e c t  speaker, have  and  resulted  without  condition,  Underarticulated  prior  the  natural  anticipation in  f o r t h e Normal c o n d i t i o n  through,  was  condition  explained  condition  the  Normal,  calculations,  and  was n o t  sentence  MI  for  the Overarticulated  conditions  Instructions  in  to  The o r d e r o f  baseline  in  (N  o f what was wanted.  nature of  conditions  Utterances  conditions  f o r m an i d e a  The  preparation  experimenters  condition  Articulated  condition  The U n d e r a r t i c u l a t e d  because the  difficult  used  nine i.e.  Overarticulated,  back t o h i s / h e r  condition,  produced  (0 c o n d i t i o n ) .  s e c o n d Normal  with  a n d an  condition,  Normally  Normal, The  speaker  articulatory  condition),  was  Underarticulated. enable  Each  and O v e r a r t i c u l a t e d  recording  i n a sound p r o o f booth  a S c u l l y model 280 t a p e r e c o r d e r  microphone.  under  condition),  recorded  further  the the  first Normal  articulatory of  the  other  articulatory  changes.  were t o s i m p l y  read the  prompting.  s p e a k e r s were a s k e d  For  the  to "exaggerate"  their  articulation  someone  with  condition,  tape  needle this  and t o  stayed within way,  the  approximately metronome keeping  a  their  timing  stop  Speakers  was  utterances  i n order  commenced,  found  Normally speakers' achieved.  to  the  If  speakers,  deflection  around  of  each  sample  that  of i t s  was  under  speaker  each  s o , most  difficult. paper  as  duration  of  times  with at l e a s t  Each  desired  The l a b e l l i n g  and  natural  three  satisfied  speakers  across  intensity  or  condition.  a s much  equal  and  speaker  before  reported  was  recording that  o f samples  they  through  as U n d e r a r t i c u l a t e d , O v e r a r t i c u l a t e d  Articulated  should  intentions  rather  t h e r e f o r e be t a k e n than  to  the  a  clarity.  two  was  kept  practiced  since their  intensity  In  Using  also  monitoring  produced  the  b u t , even  of this  speakers  was d i f f i c u l t ,  was  the  t h e 0 dB mark.  approximately  the  the  t o watch t h e VU m e t e r o f  to achieve a r t i c u l a t o r y  to rehearse  the task rest  Underarticulated  "mumble".  the  the  rates  increase  until  utterance  permitted  the  to  sentence  consecutively,  found  watch,  speaking  tendency  one  make s u r e  intensity  across conditions  Each  i f for  conditions.  a narrow range  average  as  e q u a l a c r o s s c o n d i t i o n s and s e n t e n c e s .  and  conditions.  unclear  the three  the  to  s u b j e c t s were a s k e d  recorder  clearly,  For were  still  demonstrated  addition,  very  loss".  instructions  d e s i r e d were  experimenter  the  t o "speak  hearing  the  contrasts  In  a  and  to r e f e r  condition  or  to the  actually  39  3.3 DESIGN  OF THE LISTENING TEST  3.31 PREPARATION First, utterance  the best was  hesitations, so,  OF THE LISTENING TEST TAPE one o f  selected  sentences,  of the  mainly  shortest,  or three  isolated,  based  m i s a r t i c u l a t i o n s and t i m i n g  the q u a l i t y  Generally,  and  t h e two  due  utterances while  tokens to  on  varied  to  across  intended  be  to  absence  irregularities.  unsuccessful  intended  those  tokens o f each  Even  speakers rate  of  and  control.  underarticulated  were  overarticulated  were  be  longest. Because  of  the  because  the length  so  the  that  subset  selection three  best  Test was  for  as p o s s i b l e  however,  was of  experimenters' was  included  because  his  differences, this  of  because  spite  of  productions and i t on  she  clarity Also,  The  duration  was to  her  represented  MI  own  speaker  basis  of the  as  many  One  speaker,  to  produce  or  to  (Speaker  the 1)  of h i s utterances, extremes  to discover  values  a  the  across  unable  one  limited  for  retaining  the v a r i a b i l i t y  was d e s i r a b l e the  selected  computations.  while  and  concentration,  sentence m a t e r i a l .  satisfaction. in  tokens,  n e e d e d t o be  was  utterance  i n the  the  their  tokens  speaker,  articulatory  of  test  maintain  MI  each  excluded,  variability  judgments.  the  similarity  phonemes  contrasts  could  recorded  and  conditions  quality  of the l i s t e n i n g  listeners  of the  Listening  variable  and  on  in  timing  the e f f e c t of the  listener  40  The  sentences s e l e c t e d  information Eventually, (Subject were  about three  6) and  selected,  conditions utterance  the  speakers  sentences from s i x  for  x 7  are l i s t e d  English  a total  speakers).  d u r a t i o n s and  each  o f 63 See  ranges of  is  in  Appendix  given  from  the  speakers tokens Table  in  B, Table  French (Subjects  (3  II  durations  for  I.  speaker 0 to  sentences for  and  x  listings the  SO SI S2 S3 S4 S5 S6  Table  Sex F M F F M M M  Information  Language  Dialect  Area  English English English English English English French  Western Canadian Received Pronunciation Western Canadian Newfoundland Western Canadian Western Canadian Lausanne, S w i t z e r l a n d  regarding speakers s e l e c t e d .  3 of  samples  selected.  Speaker  5)  Articulatory Speaker  Sentence  U  SO  1 2 3  2.6 2.4 2.5  SI  1 2 3  S2 .  N sec.  Condition 0  2.6 2.5 2.7  2.8 2.9 2.6  (0.2) (0.5) (0.2)  1.6 1.8 2.0  2.1 2.9 2.5  2.8 3.7 3.5  (1.2) (1.9) (1.5)  1 2 3  2.1 2.1 2.5  2.2 2.5 2.6  2.5 2.6 2.8  (0.4) (0.5) (0.3)  S3  1 2 3  2.3 2.1 2.3  2.3 2.4 2.5  2. 2 2.5 2.5  (0.1) (0.4) (0.2)  S4  1 2 3  2.2 2.3 2.5  2.3 2.3 2. 3  2.4 2.6 2.4  (0.2) (0.3) (0.2)  S5  1 2 3  2.7 2.6 2.9  3.0 2.5 2.4  2.5 2.9 2.7  (0.5) (0.4 ) (0.5)  S6*  1 2 3  1.9 1.9 1.9  1.9 1.9 2.0  2.0 2.0 2.0  (0.1) (0.1) (0.1)  * The d u r a t i o n s l i s t e d f o r Speaker 6 are f o r icorresponding French s e n t e n c e s .  Table I I . D u r a t i o n i n seconds of the u t t e r a n c e s s e l e c t e d , p a r e n t h e s e s , d u r a t i o n d i f f e r e n c e s ( i n seconds) between s h o r t e s t and longest token, f o r each set of t h r e e .  42  Once token  the s e l e c t i o n  was  copied  pseudo-random by a t  utterance  was  utterance  by  2.5 of  samples and  recorded  by  the  simultaneously, identification any  sample on To  test of  each  item  number  i f  the  each u t t e r a n c e ) ,  and  f o u r dummy i t e m s  but  no  complete  being  beginning of  each  of  each  the  next  utterance  resulted  was  i n about  satisfactory  in a pilot  test.  order could Channel for  samples  be  be  checked,  the  of the  two  track  test  item  were  was  played  on C h a n n e l  to play  could  1  each  tape  to  the  1 were h e a r d ,  both Channel  1  heard together  the experimenters wished  listening  test  (3 s e n t e n c e s x 3 c o n d i t i o n s  items  duplicate  in  and  with to  2  its  identify  the tape.  summarize,  items  tape  sample,  When t h e  the recorder  test  beginning  this  selected  utterances  numbers  speech  The  each  successive  r e c o r d e d on  2.  its  Since  the r e c o r d i n g  on C h a n n e l  setting  the  i n length,  f o u n d t o be  were  only  from  between  identification  listeners, but  was  and  other item.  seconds  silence  listening  item  one  completed,  the  seconds.  order that  speech tape,  least  separated  an amount w h i c h In  onto  each  nine  approximately seconds  twice  order,  separated  6.5  p r o c e s s was  plus (with  recorded) at  t a p e was  20.4  ten  practice  tape c o n s i s t e d  minutes.  126  x 7 speakers x 2 tokens items a t the b e g i n n i n g  r e s p o n s e s p a c e s on t h e end.  of  The  t h e answer s h e e t  r u n n i n g time of the  43  3.32 LISTENERS Ten  English  female  -  were  listeners tested male  speaking used  beforehand  undetected  using  remaining  eight  had  knowledge  t o good  3.33 PROCEDURES The  test  seven in end  some  opportunity these  items,  a n d was  was  Table  a  French  previously  replaced i n the study. of  French,  ranging III  from  provides  but  the  elementary information  test.  was l a b e l e d high  to stop  the tape  or  part  any  the tape  were e n c o u r a g e d stopping, i f the  to rate  each  HD  420  u t t e r a n c e on a sheet  instructions.  i s given  The low  (left)  end " O v e r a r t i c u l a t e d " . provided, the l i s t e n e r s  After had t h e  r e c o r d e r and a s k q u e s t i o n s of  the  was rewound  test,  possible.  i f  they  to the beginning,  to continue  tape  Sennheiser  " U n d e r a r t i c u l a t e d " , t h e mid p o i n t  (right)  h a d been  over  o f the response  as t h e  items  of  No  have  i n the perceptual  were a s k e d  listeners  end  thus  presented  this,  the  to  no k n o w l e d g e  used  Following  without  found  An example  and the  practice  Test.  five  One  was  as w e l l  of the scale  "Normal",  hearing of  and  FOR THE LISTENING TEST  point scale. C,  The  some k n o w l e d g e ,  Listeners  Appendix  Listening  fluency.  tape  headphones.  the  male  standard audiometric procedures.  had  the l i s t e n e r s  five  was  listener  listeners  -  a l l listeners  hearing loss,  of the  regarding  for  were i n c l u d e d .  candidate  Two  listeners  through  were  aimed  wished. and the  the e n t i r e  The "dummy" i t e m s at  about  test  provided at avoiding  44  Subject Ll L2 L3 L4 L5 L6 L7 L8 L9 L10 Table I I I .  any  26 25 36 26 25 33 27 29 26 22  They  such as  r u s h i n g through  consisted of  i n anticipation  items numbered  of  on the answer  presented on the tape.  THE MODULATION INDEX  3.41  DESCRIPTION OF THE INDEX  A  program was developed t o compute a measure of amplitude  modulation samples.  depth i n the d i g i t i z e d The program i s l i s t e d  Figure in  NONE SOME SOME SOME SOME SOME SOME SOME NONE SOME  I n f o r m a t i o n r e g a r d i n g the l i s t e n e r s .  sheet which were not a c t u a l l y  3.4  Knowledge o f French  Age (years)  M F F F F F M M M M  end e f f e c t s ,  finishing.  Sex  7 illustrates  a speech s i g n a l  peaks  (a's)  i n Appendix  the t y p i c a l  envelope. and  envelopes  of  the  D.  peaks and troughs found  The program i d e n t i f i e s  troughs  speech  (b's)  of  the  f i r s t the waveform  Figure 7.  Labelling of peaks and troughs i n the amplitude envelope of a speech sample.  tn  46  envelope. is  The a m p l i t u d e  then s t o r e d .  In  lowest  trough  average  (av = ( a  The  (b ; w  ratios  calculated. basic  as  amplitude  of  clarity),  n  /.av a  =  n  as  ) and their  amplitudes  i n t h e sample,  increases  normalize  (i.e.  ratio  MI  formula  t o be since  greater  decreases average  values  are  for  and  of  this  tokens  of  f o r the c a l c u l a t i o n o f  i s thus: Jo^  #  a  t  i  / a v b j  I V  b,  b,.  b„  a  a  a  1  x  b . n  t  a^  3  _ bo_ a„  a  squaring of  root  o f both  i  a  terms  sides  J  b yy,,-^  the  , ' » • • —j-*  a  j  a  »<i  a  t  "  i s t h e n e l i m i n a t e d by t a k i n g  of the equation to y i e l d  /a ay MI = n / v bi a, ' n is  i  b*.i b V "  bl  —£— •  — '  The  +  m f t K  simply  MI*  where  (a  well  The g e o m e t r i c  The b a s i c  depth  v  depth  trough-to-peak  to  lengths.  =  more  as  modulation  decreases.  taken  modulation  or  peak  trough-to-adjacent-peak  modulation  is  MI  peak/trough  b„,,„)/2).  of  t h e MI  different  the highest  a r e determined,  measure o f a m p l i t u d e  articulatory  product  )  of each  The p r o d u c t o f t h e o b t a i n e d v a l u e s i s t a k e n  the  therefore  location  addition,  rt  +  m w  and t h e  b.  t h e MI:  bn-i bn-i  a,  number o f p e a k s , av  )/2, t h e a  finally  the square  \s  is  a  value equal  a r e t h e peak v a l u e s , and t h e b ' s  t o (a^ax are  the trough v a l u e s .  3.42 The  DIGITIZATION OF TOKENS FOR MI use o f  various versions  CALCULATIONS  of the  digitized  envelope  47  signal  was i n v e s t i g a t e d .  Figure  8.  filtered some  After with  The  rectification,  cutoff frequencies  cases,  this  smoothing  amplification;  in  amplification.  Eventually,  smoothing  of the  filtering  followed  that  signal on at  others,  i n the  by  by  followed  t h e method r e s u l t i n g chosen -  i . e . 75  amplification. to a  digitized  envelopes  using  was  low-pass In  logarithmic by  linear  i n the l e a s t Hz  low-pass  I t was  reasoned  minimum, l o s s  of amplitude  w o u l d be a v o i d e d .  The  PDP-12 c o m p u t e r a t 200 Hz and s t o r e d  a s e t o f p r o g r a m s d e v e l o p e d by L l o y d  Rice  UCLA. Once  stored,  the  signals  oscilloscope  screen,  form.  s i g n a l was  and  Each  end p o i n t s  point of  followed  i t was  linear  on a  signal  o f e i t h e r 25 Hz o r 75 Hz.  was k e p t  was s a m p l e d  LINC t a p e ,  the  was  e n v e l o p e was  i f smoothing  modulation  b a s i c method i s i l l u s t r a t e d i n  the utterance, slope  Trough because cannot  wave  computation  and t h e  o f the l a s t  peak  with  them.  form o r  on  i n numerical and s t a r t i n g  were c h o s e n .  The s t a r t i n g  slope  o f the f i r s t  negative  values  implicit  Similarly,  h a d t o be  avoided  i n t h e t h e MI  formula  a trough  amplitude  of zero  undesirable  since  i t would r e s u l t  i n a c a l c u l a t e d MI  of  zero.  these  reasons,  digitized  manipulated digitized  through trough  a  peak  (see Figure 9 ) .  is  For  the  end p o i n t c h o s e n was a l w a y s on t h e  the g e o m e t r i c a v e r a g i n g with  displayed  individually,  a l w a y s on t h e r i s i n g  amplitudes  deal  be  (graphic)  inspected  f o r MI  c h o s e n was  falling  in  could  each  program  so  amplitudes  that  i t  value  envelope contained  which  was no were  FULL SPEECH SIGNAL  REVOX TAPE RECORDER  r  DIGITIZED ENVELOPE  FULL WAVE RECTIFIER  Figure  J  8.  LOW-PASS FILTER 75 Hz  9  LINEAR AMPLIFIER  Speech sample  digitization  k LOW-PASS  w  FILTER 80 Hz  scheme.  v  W  A/D CONVERTER  w  PDP-12 COMPUTER  Figure 9.  An example of start and end point locations chosen for Modulation Index analysis of the amplitude envelope of a speech sample.  50  negative  or  intensities despite  equal of  adjusted  all MI  to  upward  amplitudes for  the  efforts  instructions  of  to  zero. samples  to the or  all  At were  ensure  its  peaks  same  found  each  in  vary  through  digitized  such  had  time, to  uniformity  speakers, downward  the  a  way  the  somewhat, appropriate  envelope  that  approximately  since  the  the  was  average  same  value  for  those  utterances. values  utterances  were  selected  obtained for  the  and  listening  analyzed test.  only  51  CHAPTER FOUR RESULTS  4.1 RESULTS OF THE LISTENING TEST 4.11  CONSISTENCY OF LISTENER  Listeners the  speech  scale,  The  samples.  from  sheet  were a s k e d  1  given  data  also  since  They  between  data could provided  their  gave  their  i s illustrated  the  MI  values  t h e r e f o r e be  a check  intentions  judgments  The  computed  and  C. the  The p e r c e p t u a l  the speakers  were n o t n e c e s s a r i l y  answer  i n Appendix  checked.  o f how w e l l  c l a r i t y of  on an i n t e g e r  7 (Overarticulated).  the l i s t e n e r s  correlation  perceptual  t o judge the a r t i c u l a t o r y  (Mumbled) t o  to  JUDGMENTS  performed,  realized  i n every  case. Each the 18  speaker's performance  listeners judgments  speaker  about  produced  conditions),  six  judgments The  (N),  Each  listener  speaker's productions of  token  intended  9  tokens  was  (3  presented  were g r o u p e d ,  since  sentences twice  three  how made each x  to  according  by t h e s p e a k e r s , i n t o  deviations  p r o d u c e d by  - i.e. and  each  by a n a l y z i n g  3  the  to the sets of  each.  a function  speaker  total  utterances.  18 judgments  standard  utterances as  a  These  conditions  his/her a given  and  listeners. three  judged  was e v a l u a t e d  of the  Speakers  of 0  articulatory  Underarticulated  Overarticulated  listener  for  t o 6 a r e shown i n T a b l e IV condition  (U),  (0).  judgments  Ten  i n t e n d e d by t h e  Normally listeners  Articulated made  two  judgments number  f o r each  of sixty  listener were  judgments p e r speaker  judgments o f  the least  variable.  utterances of  from  Speaker  by  Speaker  by L i s t e n e r  judgments  by t h e  deviations  of  that  For this  6  found  t o judge  reason,  than  the data  further analyses.  6 were c o n s i d e r a b l y  other l i s t e n e r s .  conditions  The  2 and  listeners  this  who made t h e j u d g m e n t s ,  judgments  total  5 were t h e most  of standard deviations,  of the l i s t e n e r  judgments  Speakers  5 more d i f f i c u l t  other speakers.  analysis  from  indicate  fora  and p e r c o n d i t i o n .  and those o f Speaker  5 was e x c l u d e d f r o m  similar  articulatory  samples  results  produced  the  function  speech  variable,  These  utterances  A  o f three sentences per speaker,  revealed  that  consistent  than  T a b l e V shows  f o r each  intended  less  listener  by t h e  time as a  the standard across  speakers  the  (Speaker  5  excluded). In  addition,  the  was  e v a l u a t e d by  for  repeated items  the  analyses of  function Table  judgments.  as  analyzing  (hereafter  speaker a  "repeatability").  Further  of  exclusions  n o t n e c e s s a r y on t h e b a s i s  are  whose  function  of the l i s t e n e r s  the consistency of t h e i r  repeatability  of the VII  performance  shown  in  themselves judgments  The r e s u l t s o f  Table  VI  u t t e r a n c e s were j u d g e d , the of  o f these  listener  speakers results.  who  as  a  and i n  made  the  or l i s t e n e r s  were  53  Speaker  0 1 2 3 4 5 6  Underarticulated s.d. (n=60) 0.58 0.83 0.41 0.56 0.67 0.73 0.53  Normal articulation s.d. (n=60) 0.86 0.70 0.72 0.76 0.61 0.87 0.51  Overarticulated s.d. (n=60) 0.72 0.60 0.59 0.91 1.02 0.96 0.64  Mean s . d . (combined conditions) (n=180) 0.72 0.71 0.57 0.74 0.77 0.85 0.56  Table IV. Standard deviations f o r l i s t e n e r judgments a c r o s s the articulatory conditions i n t e n d e d by the speakers f o r Speakers 0 to 6 inclusive. (The d a t a were drawn from 10 l i s t e n e r s , a n d two j u d g m e n t s p e r l i s t e n e r p e r s e n t e n c e . )  Underarticulated s.d. L i s t e n e r (n=42) 1 2 3 4 5 6 7 8 9 10  0.54 0.47 0.81 0.58 0.44 0.79 0.61 0.75 0.63 0. 57  Normal articulation s.d. (n=42) 0. 59 0.84 0.60 0.68 0.23 1. 18 0.73 0.67 0.90 0.73  Overarticulated s.d. (n=42) 0.64 0.92 0.63 0.63 0.54 1 . 18 0.90 0.81 0.67 0.85  Mean s . d . (combined conditions) (n=126) 0.59 0. 74 0.68 0.63 0.40 1.05 0.75 0. 74 0.73 0.72  Table V. Standard deviations for listener judgments by t e n listeners across the a r t i c u l a t o r y c o n d i t i o n s i n t e n d e d by t h e speakers. (The d a t a a r e drawn f r o m S p e a k e r s 0 t o 6 i n c l u s i v e , and f r o m two judgments p e r s e n t e n c e f o r t h r e e s e n t e n c e s p e r speaker.)  54  Speaker  0 1 2 3 4 6  Number o f d i f f e r e n c e s g r e a t e r than 1 between r e p e a t e d judgments  9 8 4 5 9 11  Percent d i f f e r e n c e s g r e a t e r than 1 between r e p e a t e d judgments (n=81) 11.1 9.9 4.9 6.2 11.1 13.6  Table VI. Repeatability of l i s t e n e r judgments as a f u n c t i o n of the speaker. (The d a t a a r e drawn from t h r e e a r t i c u l a t o r y conditions, three sentences per speaker, and n i n e l i s t e n e r s L i s t e n e r 6 excluded.)  Listener  1 2 3 4 5 7 8 9 10  Number o f d i f f e r e n c e s g r e a t e r than 1 between r e p e a t e d judgments  1 8 5 6 1 3 6 10 10  Percent d i f f e r e n c e s g r e a t e r than 1 between r e p e a t e d judgments (n=54) 1.9 14.8 9.3 11.1 1.9 5.6 11.1 18.5 18.5  Table VII. Repeatability of listener judgments across listeners. (The d a t a a r e drawn from three articulatory conditions, and t h r e e sentences f o r each o f s i x speakers S p e a k e r s 0, 1, 2, 3, 4, a n d 6.)  55  For of  the  all  other  two judgment s c o r e s  token r e p l a c e d the  4.12  analyses  of  the  perceptual data,  from each o f  individual  the  listeners  the  mean  for  each  scores.  COMPARISON OF L I S T E N E R ' S  JUDGMENTS WITH SPEAKER'S  INTENTIONS In  Figure  separately  for  articulatory are  separately. be s e e n  for  every  to  of  the  agreement  for others. the  best  For  the  N vs.  speakers listeners.  each  with  the  f o r some  Speakers  the  three-way there  for  i n each  condition  apparent.  and  listener listener  intentions considered,  speakers'  speakers  data  each  speakers'  intentions In a d d i t i o n ,  and  listener's  some  and L i s t e n e r s  162  of  listeners 1,  3,  and 5  sets  articulatory conditions ( U of  judgments,  p r o d u c e d 3 s e n t e n c e s w h i c h were The  of  the  respect.  contrast were  means  intentions  2 and 6,  in this  The same  s e n t e n c e s were  between  speaker's  subjects  O),  well  scores,  a function of  judgment s c o r e s  p e r c e p t i o n became  was b e t t e r  the  pooled,  When i n d i v i d u a l  between  judgment  speaker.  to d i s p l a y  the  correlate  listeners'  perception  listener  a r e p l o t t e d as  speakers  some d i s a g r e e m e n t  the  vs.  11  all  speaker.  and  were  Figure  The means  can  of  c o n d i t i o n i n t e n d e d by t h e  scores,  however,  t h e means  each s p e a k e r ,  arranged in  judgment  than  10,  listeners'  perception  each agreed  since  6  j u d g e d by 9 with  the  Listener  Listener  Judgments -  P(JO  >  c  fD  3  rr  C  CO 15 fD CD  M  ?r fD i-l  rr fD  3 a.  H-  rr PO  CO rr (D B f0 "1  CO  (_..  < Pt>5  3  •  C CL  O |-C cr* << •< CO TJ CD 03  7? ft) i-i  09  tsi  CJ  Judgments .p-  Kj\  CN  cn •a  > i-i  fD CU  rr P-  fD i-l  n c  3 P> a . Cu rr O i-l  Z  1-4  cr v :  3  p. rr p. O  rr fD  CL  o o  CL  3  fD  fD  CL  P-  1  O  CO XJ fD 0)  M  O O  3 a. P-  ?r r r fD i-i  3  PO  3  3  (T>  3  rr CO  (»  CO  Listener  to  Listener  Judgments  fu  H  a  or r  PO  3  of rr  3"  CD CO  •a CD m  CD  •  P-  3  T3  fD CU 7? fD I-i  m 3  CL fD  CL O  t» Ptn i-i  3  rr fD  3  cn  > i-i rr P* O  TJ  fD CD 7? fD  C  C M  CU fD r r CL O •I  to  o z cr >< n tn o •o s fD  P-  h-  CL  a . i-i  7?  o  CO  > i-i rr P-  Judgments  n  M  Z  cr CO  •O  Cl  o  3  fD C L 03 p . TT rt fD P •1 O  n  rt p-  o  3  O  3  II  3  n  CD  3  <_>.  Listener  c  Listener  Judgments  Judgments  ClCW  3 tn 3 rr  P-  3  rr fD  > i-i rr  cn  TJ  H-  fD CU  C  fD  c  ?c  n  l-l  3 t-i a . cu fD  CL  c o a  a  cu n a.  99  cr •< co  -O fD CB IT fD i-l  rr O  ON Z  Prr PO  3  rr fD  3  CL  fD  cr  n o CL  3  CL  +  H  3  P-  I-i  fD  3  O  rf  O1  3  co  TJ  fD 0> K fD •I  cn  >  TJ  rr C PO  (D »  tfl  fD r-i  C  P< 01 rr O I-I  Z  r H  o o a  CL  p. rt pO  a  O  V-0-«  Listener Judgments  Listener Judgments  Ul  Ul  OO  c n  re r* *  CL  ft  rr  < rr u 3re rt CO  re  HO n  3  cn •  c e. 05 s  >  H* ft 3 H" n  C  re c 3 — f' CL Qi  3  CT  n  CL  re 3  n  o 3  re  O  Z  vj  cn "O  CU  ?r PT re Hi-l  I-  rr CL O i-<  p( cr vs ><  (-u  re re  3  (B  PT CL O Z  CU  PT  C  CL 01  re  o 3  cn  re 3 re ro c  PT  cn •a re  f  > i-l H* rt 3 HPT n  o  CL  H-  JC PT H-  re n  o 3  o  3  r cn /—V  (u cn  o o p PT  l-Ti  C  n  Listener Judgments  Listener Judgments  rO  C  3  U)  rePT H-  a. 0  *  •  a *—'  >  3 O  Ml  PT  3*  re rH* cn 1  PT  re 3 re p(  •  r  3  PT  n Hn  C  re 3 re  re c 3  i-l  r-  a - cu re PT CL O II  PT  CL  PT  re  P)  3  CP  re •-I  I-  Hi  rr o  a  n  cn  C  re c ro  •OH  Z  o 3 re CL o +t 7T re n  H* rt H-  3 3  O" vcn  rr  >  rt  ro  z  v:  n  ca o •a p re a. o cu p. jr PT re p* 11 o  rt  H-  o  3  3  5 re 3  ro  3 re  1  3  PT  >  M H- rt 3 Prr n  C  re c 3  M  O. CU ft) rr CL  o CU  3  CL  CD i-l  Listener Judgments  Listener Judgments  C  a co  o z  II  cr v o cn o 3 re CL o cu p?r PT ro H n o P  UJ  1  -C-  1  Ui  1  CT\  1  r-  «J  h rr Pcn rt ro  3  ro ii  H  > H* rt 3 P-  Z  cr *<  cu ?r  ro n  Ui  r-  3  3  o cn o •a 3 roCL  ro  cn rt re  C  ro  a ro  1  O  M  ro »i u>  Listener Judgments J  ro  UJ  .ts  i_n  CTN  »j  > a  H-  rr  f>  (X  01 rt  f  c  01 rr  ro C a t-  a  CD  i-i  CL  O  01 0)  Oi ?r (t> r<  CO  z  cr vs vj TS  rt>  ro  oO a  CL  p. rr o P0  a  Listener Judgments N5  Ul  -C»  -r-  U U ll  Ov  -I—  >  Ul rr ro 3 ro ii  w* rt 3 Vrt n  c  n> r3 Cu 0> rt rt O.  cr vs  1  O  ii  01 •O  o o3  ?r  pt  i-l  o a  z  ro CL 01 H- o ro  H-  Listener Judgments ts>  _,  > ii 3 rr  rt H* O  CL  CL ro Qi H-  f c rr  on  roH-  -t>  1  Ul  CT>  ,  -»J r 1  c  HCO rt ro 3  ro  w  H cr v ; vs O 01 O •a 3  1  r  ro 3 c rt ro CL O  UJ  r| z  o  59  speaker's the  intentions i n  two-way c o n t r a s t s  (135/162) there  forN  was, i n  speakers'  was 88%  v s . 0,  the  the  speakers  intentions  (9  listeners x  between  N,  83% Thus,  between the  perception  of  JUDGMENTS  the e f f e c t s  and the  o f language  on  the  p e r c e p t u a l data was s l i g h t l y  3 sentences)  the speaker's  i n t e n t i o n s and  articulatory  conditions.  (99/135 judgments)  The  their  utterances French speaker's knowledge  l i s t e n e r s f o r a l l three  agreement between speakers'  the l i s t e n e r s  judgments.  in  100%  intentions, o f French  11 of  by the  the r e l a t i v e ranking of  For the E n g l i s h speakers,  are considered,  were  sentences,  there was intentions  judgments.  f a m i l i a r i t y of  affect  Of the 27 s e t s  17 (63%) had complete agreement  c l a r i t y assigned  listeners'  speech.  a p p l y i n g to the three French  articulatory  only  listeners'  f o r the E n g l i s h than the French  l i s t e n e r judgments  to  vs  clarity.  drawn about  of  and  for U  However, based on the data a v a i l a b l e , agreement between  better  73%  Agreement f o r  small s i z e o f the corpus used l i m i t e d the c o n c l u s i o n s  c o u l d be  data.  (143/162)  EFFECT OF LANGUAGE ON LISTENER  The  sets.  monotonic r e l a t i o n s h i p  and  increasing a r t i c u l a t o r y  which  o f 162  and 99% (161/162) f o r U v s . 0.  general, a  intentions  4.13  117 out  21 sentences  In f a c t , the  (6/6  two  when only listeners  sentences)  whereas  agreed  with French  with  (52%).  the  7  the who  agreement listeners  the speaker's  a l s o seems French  knew with with  no the some  intentions for  Perhaps the l i s t e n e r s with no  60 French the  were able  speaker than  latter the  group may  to c o n c e n t r a t e the  not  listeners  with some  of the sentences.  allow s t a t i s t i c a l  French, s i n c e  The  small  the  sample s i z e  a n a l y s i s of the d i f f e r e n c e  or p u r s u i t of t h i s p o s s i b l e e x p l a n a t i o n ,  4.14  the a r t i c u l a t i o n of  have been d i s t r a c t e d by attempts to d e c i p h e r  semantic content  does  more on  observed  however.  EFFECT OF UTTERANCE DURATION VARIABILITY  ON  LISTENER JUDGMENTS Speaker Test  1 was  because he,  contrasts  of a l l  of u t t e r a n c e  conditions. agreement  the  Various  digitizing  the  MI  agreement  f o r the  a cutoff  for  (p = 0.05,  (logarithmic  vs.  linear)  not  the  other  z =  1.55  75 Hz,  v a r i a t i o n s explored, greatest  amount  I t was  75  Hz)  were e x p l o r e d  in  computation.  f i n a l measurements  d i g i t i z e d sample.  was  CALCULATIONS  frequency of  of the  81%)  Test).  the data i n view of MI  a l l the  or  combinations of low-pass f i l t e r i n g  amplification  retention  the  speakers (77/108 judgments or 71%)  THE  Of  speaker/listener  judgments  RESULTS OF  greatest  three a r t i c u l a t o r y  1 (22/27 from  Listening  a c r o s s the  f o r Speaker different  the  produced the  shows that  4.2  with  speakers,  a t-test  Proportions  selected  inclusion in  duration  on a t w o - t a i l e d  and  for  However,  significantly English  selected  (25 Hz,  The  were low-pass and  parameters filtering  linear amplification.  t h i s combination r e s u l t e d i n of amplitude  modulation i n  reasoned that s e t t i n g the  cutoff  frequency  of the  f i l t e r as  limitations)  and  amplification  should a v o i d  amplitude  modulation  otherwise  MI  A  decreasing  signal. that  value.  intended by  the  order UNO  its  l a b e l according  the  three tokens  for  values. the U  t h i s a n a l y s i s are l i s t e d i n i n t e n t i o n s f o r each  i n t o MI  Consequently,  amplitude MI  modulation  order of  the speaker  f o r each sentenoe. to speaker  per sentence  For example,  in  ranking i s a s s i g n e d  The  result  envelope.  ranking  s m a l l e r MI value i n d i c a t e s more  clarity, MI  logarithmic  d i f f e r e n t samples which might  o b t a i n e d with  i n the  articulatory  MI  than  l o s s of c o n t r a s t i n amounts of  The MI values were arranged  hypothesis  clarity  for  any  rather  f u n c t i o n of the speaker's  Table I X ) .  modulation the  values  V I I I as a  utterance. (see  linear  from e x c e s s i v e smoothing of the  The Table  using  high as p o s s i b l e (given equipment  was  orders  amplitude  accordance  with  increases  with  as a f u n c t i o n of  increasing  articulatory  assumed to oerrtiipend to  Eaoh of the tokens r e t a i n e d i n t e n t i o n s , but  c o u l d be rearranged  a sentence  with MI  the order of a c c o r d i n g to  values of 0.789  c o n d i t i o n , 0.876 f o r the N c o n d i t i o n , and 0.688 f o r  the O would be a s s i g n e d an MI  ranking of  As can be seen i n Table IX, the MI  NUO. ranking order  failed  Modulation Speaker Sentence  Underarticulated  Index  Values  Normal articulation  Overarticulated  0  1 2 3  0.819 0.800 0.936  0.906 0.884 0.865  0.923 0.926 0.832  1  1 2 3  0.886 0.860 0.871  0.848 0.872 0.847  0.897 0.868 0.902  2  1 2 3  0.767 0.885 0.779  0.586 0.813 0.808  0.683 0.842 0.828  3  1 2 3  0.894 0.896 0.860  0.762 0.885 0.844  0.860 0.909 0. 788  .4  1 2 3  0.821 0.889 0.831  0.674 0.822 0.795  0.699 0.833 0.791  6  1 2 3  0.842 0.923 0.839  0.875 0.916 0.868  0. 778 0.752 0.751  Table VIII. M o d u l a t i o n Index v a l u e s f o r u t t e r a n c e s by s i x speakers. (The d i v i s i o n o f v a l u e s by a r t i c u l a t o r y c o n d i t i o n r e f l e c t s the i n t e n t i o n s of the speakers.)  63  Speaker  Sentence  Modulation ranking  0  1 2 3  ONU ONU UNO  1  1 2 3  OUN NOU OUN  2  1 2 3  UON UON ONU  3  1 2 3  UON OUN UNO  4  1 2 3  UON UON UNO  6  1 2 3  NUO UNO NUO  Index  Table IX. Modulation Index rankings of u t t e r a n c e s by articulatory condition. (The labels for each token corresponds to the a r t i c u l a t o r y condition intended by the speaker. According to the experimental h y p o t h e s i s , a l l MI ranking orders would be UNO i f the speakers were completely s u c c e s s f u l at producing the d e s i r e d a r t i c u l a t o r y c o n t r a s t s . )  64  to  match the  between  MI r a n k i n g  sentences)  For  three-way  the  Agreement  was  (U v s .  N  22% vs.  (4/18 0)  the three two-way c o n t r a s t s , U v s . N., 72%  (13/18), 39%  was N vs.  (7/18), and  61%  respectively.  Sample of  intentions  contrast  and U v s . 0, agreement was  (11/18),  consistently.  and speakers'  when a  considered. 0,  speakers' i n t e n t i o n s  s i z e s were  effects  of  too s m a l l to a l l o w meaningful language  or  of  utterance  analysis duration  v a r i a b i l t i y on MI v a l u e s .  4.3  COMPARISON OF MI VALUES WITH PERCEPTUAL DATA  4.31  COMPARISON OF MI VALUES WITH LISTENER JUDGMENTS  MI  values ranking  o r d e r s were  compared to  judgments  ranking o r d e r s .  according  to the speaker's i n t e n t i o n s .  comparisons, of  the s i x  sentences was  f o r which  listeners  there was  the  analysis  In  way  of  selecting  the  most agreement between l i s t e n e r s  were  those  for  shown agreement  or sentences  judged two  There were 18 p o s s i b l e  sentences out of the o r i g i n a l eighteen  had p r e v i o u s l y  intentions,  clarity.  However, some  The o n l y  in  retained i t s label  each of three sentences produced by each  speakers.  needed.  included  had  one f o r  Each token s t i l l  the l i s t e n e r  f o r which  which  a l l nine  with the  speakers'  one l i s t e n e r  out of nine  tokens to have the same degree of a r t i c u l a t o r y other words,  at worst,  one l i s t e n e r had  judged  as e q u i v a l e n t two d i f f e r e n t tokens of the same sentence. Five 3  sentences  produced by  (1  Speaker  each produced by Speakers 0 and 2)  met  the above c r i t e r i a  1, and  i n a l l three  conditions. N  v s . 0,  on  This  a perceptual  the b a s i s of  None  of the  the  five  agreement). vs.  0,  was  and  61%  perceptual data produced  between  order  o f UNO  c o u l d be  judgments from a t l e a s t  sentences  agreement  speakers  ranking  had an MI  ranking  assigned  eight  listeners.  order  o f UNO  (0%  F o r t h e two-way c o n t r a s t s U v s . N, N v s . 0, and U  sentences), the  meant t h a t , i n t h e t h r e e - w a y c o n t r a s t U v s .  the U  71% (11/18  and  sentences),  sentences),  the  more  and N  (10/14  MI  27%  (3/11  respectively.  Both  values  indicate  in  articulatory  contrast  c o n d i t i o n s than  between  that  the  the  clarity N  and  0  conditions. In MI  order  values  (Speaker best  to  S p e a k e r 6)  listeners  trends  relationship  great  with  t o be  t o be  relationship particularly  lower  over-  scatter between  strong.  1,  were  normally  deal of  was  (Listeners  Unexpectedly,  judged  t h e r e l a t i o n s h i p between t h e  and t h e p e r c e p t u a l d a t a , 2 and  listeners  better evaluate  or  data  plotted 3,  and  MI  the  the best  speakers  against data  from t h e  5)  suggestive values  (see of  for  underarticulated  articulated. in  from  these MI  and  There  plots, this  a  Figure  12).  non-monotonic  tokens  judged  than  for  was,  however,  indicating perceptual  tokens  that data  by  a the  i s not  11  0.5 "  n  0.4  0.3 . i  M  u  o o  u  0.2  o o ti  n  0.1  n  0.0 2  3  4  5  Listener Judgments - Listener 1  Figure 12a.  The relationship between Modulation Index values and l i s t e n e r judgments. MI values are for utterances by Speakers 2 and 6. u indicates the point corresponds to an utterance intended by the speaker to be Underarticulated. n indicates the point corresponds to an utterance intended by the speaker to be Normally A r t i c u l a t e d , o indicates the point corresponds to an utterance intended by the speaker to be Overarticulated.  o  0.5I  n  u  u  o o o un  0  n  Listener Judgments - Listener 3  Figure 12 b.  0.3  4o y n  o o o  B n  : 1  • 2  1  3  • 4  H — 5  L i s t e n e r Judgments - L i s t e n e r  Figure  12c.  5  0.5  n  0.4.-  «  n  0-3  0.2 ••  0.1  0.0 2  3  4  5  -V-  4-  6  7  Listener Judgments - Listeners 1, 3 & 5 combined  Figure  12d.  The line indicates the non-monotonic relationship suggested by the points.  70 4.32  COMPARISON OF MI VALUES WITH VISUAL IDENTIFICATION OF WAVEFORMS  In  an  visually or  samples,  but  orders  the  based  comparison,  U  vs.  on  c o n d i t i o n s (U, N , selected  oscilloscope  identifications out  of  that  vs.  O  30  the  after matched  trials.  A  author  contrast.  compared to  the  a confusion  listeners' matrix  judged to  by the l i s t e n e r s  could  The  the p e r c e p t u a l ranking  judgments.  (Table  have low  visual  For  this  XI) again  showed that  or  articulatory  high  were those v i s u a l l y  i d e n t i f i e d by the  as U or O r e s p e c t i v e l y , whereas tokens which were r a t e d  the middle  listeners  o f the  Three waveforms  from the  comparing  clarity  by  the  as N.  three  observed i n articulatory  tokens, as  modulation,  articulatory  i d e n t i f i e d as O most o f t e n , as U some  d i f f e r e n c e s were  Overarticulated of  scale of  were v i s u a l l y  the time, but seldom  depth  reveals  N  were a l s o  which were  clarity  of  X)  11  to  d i s c e r n the U v s . 0 c o n t r a s t intended by the speakers,  identifications  author  i n only  attempted  randomly  on an  The v i s u a l  matrix (Table  not  tokens  author  various  were d i s p l a y e d  speaker i n t e n t i o n s  readily  the  of the  equalization.  confusion  test,  intended a r t i c u l a t o r y  the waveforms  when they  amplitude  in  blind  i d e n t i f y the  O) from  the  informal  which  the  shapes  conditions.  hypothesized, presented could  be  clearly  of the First, the  observed  the e x c u r s i o n of adjacent peaks and troughs.  most by  71  A r t i c u l a t o r y Condition visually identified  Articulatory Condition intended by the Speakers  U  N  0  U  5  5  0  N  4  1  5  0  1  4  5  Table X. Confusion matrix comparing v i s u a l i d e n t i f i c a t i o n of a r t i c u l a t o r y c o n d i t i o n with speaker intentions.  Articulatory Condition visually identified u  Articulatory Condition based on rankings of tokens a c c o r d i n g to Listener Judgments  N  0  u  55  N  16  8  30  0  10  34  62  ,  4  7  6  Table XI. Confusion matrix comparing v i s u a l i d e n t i f i c a t i o n of articulatory c o n d i t i o n with the a r t i c u l a t o r y condition suggested by the p e r c e p t u a l ranking order a c c o r d i n g to the l i s t e n e r s ' judgments.  Second,  whereas  rounded,  with  tokens peak  a  tended  few  to  contained  Normally  U n d e r a r t i c u l a t e d tokens main  have  peaks  jagged  several  conditions.  The  third  troughs,  peaks,  smaller  A r t i c u l a t e d tokens  and  appeared  and  had  observable  top  and  features  and  Overarticulated  the  peaks  smooth  of  each  main  troughs.  of  the  difference  The  other  two  deserves  more  discussion. The  Overarticulated  Normally where  A r t i c u l a t e d tokens,  the  time.  can  the  where the  be  seen  is  value  trough  MI  for to  troughs  the  designed  the  number  was  the  take  bottom  peaks,  av  extent,  in  their  some  short  period  calculation  of  words.  since  zero  clipping  the  envelopes  between  effect,  l i m i t e d to  formula for to  lesser  pauses  "bottom-clipping"  not  of  the  trough,  the  intensity. of  into  the  MI,  account.  a  l  of  highest l  9  n  are  peak, the  =  (a  m 8 J t  and  b  peak  values,  m i / (  is  + b^;* )/2, the  a o, i s m )  value  of  the  a n d b,^„., a r e  the  values.  Samples zero  from  a  that:  n  lowest  of  to  zero  corresponded  a  excursion  MI was  Recall  in  and  had p l a t e a u s  dropped to  plateaus  resulted  downward As  intensity  These  This  tokens,  trough values  were  manipulated prior  amplitudes  depend  w o u l d be  to  MI c a l c u l a t i o n s  eliminated.  on t r o u g h - t o - p e a k  amplitude  so  Therefore, ratios,  if  that since  trough  amplitudes  are  are  MI  not,  containing mainly  limited values  MI  values  which  i s the m o t i v a t i o n  that  this  tokens  are  elevation  artificially  for this of  MI MI  order  by  reducing  judged  by  the  ranking  to the  weakened  and  the MI  differ  for  hypothesis  the  and  between  to  apply  It is possible,  order  contrasts  tokens  would  articulated,  study.  listeners  values  according  values  amplitudes  elevated for  more c l e a r l y  expected  between t h e  when peak  e l e v a t i o n o f MI  which are  low  ranking  be  This  which  relationship  bottom c l i p p i n g  will  pauses.  to tokens  by  then,  apparent perceptual  values  in  for  articulatory  clarity. This It  was  reasoned  limited instead  possibility  of  t h a t MI  formula  peak and  following  i s the m o d i f i e d  av  = 1/2  then  contrasted resulting  would  i f the  a subset be  MI  in MI  modified values  MI  less  formula  of the  data.  affected  by  were a r i t h m e t i c  (a „ + ma  the  ranking  brrt*  i n ranking  Sentence 3),  the  the  a  of are  two  {Speaker  rather  + £(b;  for  m o d i f i c a t i o n does For  include than  differences  ratios.  The  - a - )1  ) .  number orders  to  formula:  l / n [ a v - a„  recalculated  significantly. change  was  trough  MImod =  The  values  with  geometric. MI  between  was  explored  excursion of troughs  The  MI  was  subset pauses  of  tokens  they  contained.  shown i n T a b l e  not  change  sentences 3,  s e p a r a t i o n i n the  Sentence original  The  XII.  t h e MI for  which  ranking  which  orders  there  1  and  Speaker  MI  values  is a 4,  74  Speaker  Sentence  0 2 3 4 6  1 2 1 3 1  Table  peak/trough  between  Comparison  MI  Peak/trough  MImod ranking  ONU UON UON UNO NUO  XII.  original  MI ranking  values ratios  t h e two  consequently  of  with are  differences  ONU UON UON UON NUO rank  rankings the  according  according  basis  f o r MI  to  to the MImod.  calculations;  a r e t h e b a s i s f o r MImod s c o r e s .  c o n d i t i o n s which  the  orders  changes  in  r e v e r s e d was v e r y rank  cannot  be  s m a l l , and considered  significant. There between  was an MI  articulatory  values  o f improvement  that  some o t h e r elimination  better  solution.  with  this  i n rank  method o f of  i n most c a s e s  calculated  clarity  lack  as  improvement  pauses  for  tokens  modification.  ordering of d e a l i n g with  from  i n the separation of  contrasting  Even  conditions  so, the suggests  bottom c l i p p i n g ,  the tokens  analyzed,  such  may be a  75  CHAPTER  FIVE  DISCUSSION AND  The for  estimating  affected basic the is  purpose  by  of this  the  the  CONCLUSIONS  s t u d y was  the e x p l o r a t i o n  intelligibility  articulatory  premise behind  this  of  different  clarity  computed  amplitude  by  greater  envelope  speech.  measure, t h e  MI,  less  modulation  as The  i s that  for intelligible  amplitude  for  speakers  of t h e i r  a m p l i t u d e e n v e l o p e o f t h e waveform characterized  o f a method  speech  than  intelligible  the  speech;  consequently,  a measure o f a m p l i t u d e m o d u l a t i o n s h o u l d p r o v i d e  a  speech  measure o f  previously  applied  intelligibility and  their  Houtgast,  1983). Function  1980; and  receiver compared  the  more  and  as t h e  basis  and  1980;  1984;  developed  for  idea  case  Houtgast,  Plomp, 1980;  the second  been speech  Steeneken, 1973,  198-1;  Houtgast  and  van R e i t s o h o t e ,  Steeneken the  has  of  Steeneken,  Houtgast,  1981,  authors  general  and  Plomp,  Steeneken  This  s p a c e s by  (Houtgast  Steeneken,  These  and  Houtgast  Modulation  Transfer  version  of  the  Speech  Index.  room  acoustics,  of the before  transmission  the  in listening  Steeneken  Transmission In  to  colleagues  Steeneken, Houtgast  intelligibility.  the  changes  amplitude spectrum and  system.  after  corresponding reduction  o f a speech  passage  Reduction in  between  through  source sample may a  are  be  speech  i n amplitude modulation intelligibility  and  due  and to  76 several  causes,  filtering In  effect  sum  and Steeneken,  is  8000 Hz  modulated  the  In  speaker,  natural  makes  an a v e r a g e  when  of  intensity  speech  comparing  must be  natural  used,  which  found  the present  study, a  t o produce  Overarticulated.  listeners, duration  input  o f 125  spectrum  in  natural  intelligibility  o f one  since  i t is  not  interest.  the This more  speech  are u n l i k e l y to  of  must a l s o be e q u i p p e d  effects  d e s i g n was  and  t o cope  of these other  used  in  sentences i n  ( o r mumbled),  Efforts  selection  the  t h e same  and d u r a t i o n  samples,  differences.  minimize  Underarticulated  intensity  of a  Function.  amplitude  are of  i n natural  speech  and t i m i n g  order to  attempted  Houtgast,  consisting  c o n s t a n t a c r o s s s p e a k e r s and a c r o s s speech  In  and  frequency range  frequencies  speech  Steeneken  Modulation Transfer  across the  frequencies  measure u s i n g  ways:  of the  speech  signal  and  than the task o f the Modulation T r a n s f e r F u n c t i o n .  modulation  with  the  o f t h e p r o p o s e d measure, t h e M I , somewhat  the task  remain  and  bands o f p i n k n o i s e a s an  modulation  properties  complicated The  modulated  contrast,  Steeneken  artificial  thus shaped  speech.  average  of the STI, Houtgast  a simple  t o resemble at  reverberation,  1973, 1985;  f o r measurement  signal  to  in  version  of sinusoidally  The  noise,  o f the system.  1983) u s e d  signal  a  particular  the second  (Houtgast 1980,  in  which three  the tokens  across a r t i c u l a t o r y  speakers different  Normally A r t i c u l a t e d ,  were made t o m i n i m i z e  o f tokens,  factors  through  conditions,  differences in  instruction  with l e a s t and  and  to the  variability in manipulation of  77 samples as  to equalize  intensity  One  to the  speakers  the adequacy  were  articulatory This checked were  in  this  were  more  to  articulatory listeners  more  so, i n  much  better  was  clarity  be d i f f i c u l t , others  compounded  when  the  data.  Normal  speakers'  clarity  they  perceived.  were  fairly  successful  fact, (and  than  t h e MI.  producing  in  differences,  of French  hearing  in this  were  listeners as  to  of the  the  that  r e s p e c t - much  listeners  speech which  observation that  were  were  tokens ( i . e . the  MI was n o t  listeners  with  no  a t judging the a r t i c u l a t o r y  i n t e n d e d by  a French  speaker  of French  suggests  that  o t h e r than  the  others i n guessing the  etc.)  seemed b e t t e r  by f a c t o r s  values  Results indicate  some  cues  The  MI  However, some l i s t e n e r s  Perhaps  to a d d i t i o n a l  distracted  as  a n d some  in  utterances  more c o n s i s t e n t ) t h a n  to d e t e c t .  knowledge  produced.  task of producing  than  r a t e , the  or i n t e n s i t y  knowledge  to  they  the  time and feedback  successful  intentions.  designed  remained.  contrasts.  asked  responding  such  t h e r e f o r e remained:  the type of speech  proved  against perceptual  timing  study,  the tokens  productions, the  contrasts  problem  speaker's  within  allowed rehearsal  of their  articulatory  speakers  the  error  more s p e c i f i c a l l y  Although  Even so, c o n t a m i n a n t s  and t i m i n g v a r i a t i o n s  source of  speakers,  intensities.  than  the those  listeners  listeners of  with  could  interest  to  no be the  experimenters. In  spite  relationship  of was  a l l of observed  the  difficulties  between the  encountered,  MI v a l u e s  a  (calculated  for  the u t t e r a n c e s  of the best speakers) and the judgments of  the  best l i s t e n e r s  (see F i g u r e  non-monotonic  r e l a t i o n s h i p , with  judged  Underarticulated  to be  those  12).  The t r e n d was  lower MI or  values f o r  Overarticulated  tokens judged to be Normally A r t i c u l a t e d .  this  r e l a t i o n s h i p presents  monotonic  a problem.  curve suggested  by the  toward a tokens  than  for  The nature of  A c c o r d i n g to the non-  d i s t r i b u t i o n of the p o i n t s ,  low  MI v a l u e s c o r r e l a t e with both extremely high and  extremely  low  articulatory clarity,  correlate  with  average  modifications establish  the  clarity,  intelligibility  rating  order to be, the  tokens  d i s p l a y e d on  served  to i d e n t i f y  the  values  Clearly,  required between  i s to  be  in MI  a  some  order  values  useful  to and  speech  tool.  experimenter  some of  studied  these  the  an o s c i l l o s c o p e .  modifications  envelopes  There  of  the  were cues which  the a r t i c u l a t o r y c o n d i t i o n s .  As expected,  modulation depth observed i n the tokens i n c r e a s e d with the  articulatory to  the MI  d i s c o v e r what  should  are  relationship if  MI  clarity.  measure  monotonic  articulatory  In  articulatory to  a  whereas h i g h  clarity  t h i s expected  between  words  increased.  perceived  contrast,  increased  Although  1986)  speech  intelligibility,  pauses  as  the  perceived  number of pauses  number  of  articulatory  pauses clarity  i s a f a c t o r c o n t r i b u t i n g to  f o r the purposes of MI c a l c u l a t i o n s of  modulation, the  are b r i e f  however,  In a d d i t i o n  a c c o r d i n g to Picheny, Durlach and B r a i d a  (1985,  amplitude  the  by the l i s t e n e r s .  pauses were  p e r i o d s of  silence,  contaminants.  Since  the i n t e n s i t y at pauses  79  descends  to zero  begins.  This  not  results  accounted  modification of  bottom  for  or  addition  of  A  final  dependent  to  the f i r s t  or t h i r d - o c t a v e bands  transmission  i s that  specific precision.  in  selection  pauses  f o r MI  analysis,  values  we  equipment  French  and  Index.  wide  band  band(s) of i n t e r e s t  bands,  from  to  Steinberg  signal,  analysis  may  be be  the speeoh which  was  in  the  octave  the s i g n a l  within  passage through a division s u c h as or  measured  frequencywith  more  to  the  significance  averages  the  bands.  of the  low-pass  important  the other u n a f f e c t e d  when  t o employ  noise,  but t h e i r which  of  was  (1947)  analysis  of t h i s  band can  time,  intelligibility  effects,  narrow  were u n a b l e t o and  eaoh  b e f o r e and a f t e r  effects  with  obtained  They d i v i d e d  and t o compare  modulation,  of the  without  be t h e  which  The a d v a n t a g e  These  intelligibility lost  in  frequency-specific  amplitude  effect  may  t o MI  MI  equally  selected  interfering  A  MI.  tested  f r e q u e n c y dependent  bands,  system.  the  to reduce the  A more modern a p p r o a c h has been  of the  filtering,  to the  frequency  to contribute  bands  of  which i s  pauses.  t o employ  twenty  t h e n e x t word  effect  was  solution  factor  the A r t i c u l a t i o n  into  scores  analysis.  case.  signal  MI  limitations  ideal  each  designed  modification  frequency  calculated  formula  a correction  due  spectrum  calculation  w h i c h do n o t c o n t a i n  explore,  calculating  the  A better  tokens c o n t a i n i n g  until  a "bottom c l i p p i n g "  on  results.  speech tokens  were  MI  clipping  of  remains there  in in  to the  satisfactory  speech  and t h e n  may  be  frequency  80  To  summarize,  intelligibility modulation based MI  on t h e  and  speech results  suggestions here,  need  by  of this  before i t  t h e Index w h i c h w i l l  becomes a  of  seems  study.  speech amplitude promising,  However,  useful  tool.  of inter-speaker v a r i a t i o n s  t o be  research  rating  envelopes  exploratory  overcome  for modifications  further  of  quantification  amplitude  the e f f e c t s  intensity  concept  of speakers  needs m o d i f i c a t i o n  particular,  of  in  the  will  adequately.  to the  MI have  effective.  In  i n timing  Although  some  been p r e s e n t e d  be n e c e s s a r y t o d i s c o v e r  be most  the  the  form  81  BIBLIOGRAPHY A h l s t rom, C. and. Humes, L.E. (1983). "Normal modulation transfer functions i n n o i s e u s i n g temporal probe tones," ASHA. 25(10):67. Ahlstrom, C. and Humes, L.E. (1984). "Psychoacoustic modulation t r a n s f e r f u n c t i o n s of impaired ears u s i n g probe tones," ASHA, 26(10):175. Ahlstrom, C, Boney, S.F. and Humes, L.E. (1984). "Modulation t r a n s f e r f u n c t i o n s of h e a r i n g a i d s , " ASHA 26(10):124. Ananthapadmanabha, T.V. (1982) " I n t e l l i g i b i l i t y c a r r i e d by speech source functions: implications f o r a theory o f speech p e r c e p t i o n , " Speech T r a n s m i s s i o n Laboratory, Q u a r t e r l y Report. STL-QPSR 4/1982. Ando, K. and Canter, G.J. (1969). "A study of s t r e s s i n some E n g l i s h words by deaf and h e a r i n g speakers," Lang.Speech 12:247-255.  syllabic normally  ANSI (1969). American N a t i o n a l Standard methods f o r the c a l c u l a t i o n o f the a r t i c u l a t i o n index (ANSI S3.5 1969) New York: ANSI. Black, J.W. (1957). " M u l t i p l e c h o i c e i n t e l l i g i b i l i t y J . o f Speech and Hearing D i s . 22(2):213-240.  tests,"  Brannon, J.B. (1966). "The speech p r o d u c t i o n and language o f the deaf. Lang.Speech 12:247-255.  spoken  Crum, M.A. (1974). " E f f e c t s o f speaker to l i s t e n e r d i s t a n c e upon speech intelligibility in r e v e r b e r a t i o n and noise," D o c t o r a l d i s s e r t i o n . Northwestern U n i v e r s i t y . Dareham, J.R. (1986). "Measuring speech intelligibility u s i n g RASTI," Sound Video C o n t r a c t o r 3(11)50,52,54,5657. D i r k s , D.D., B e l l , T.S., Rossman, R.N. and K i n c a i d , G.E. (1986). " A r t i c u l a t i o n index p r e d i c t i o n s o f c o n t e x t u a l l y independent words," J.Acoust.Soc.of America 80:82-92. Doyle, J . (1987). " R e l i a b i l i t y o f a u d i o l o g i s t s ' r a t i n g s of the intelligibility o f h e a r i n g impaired children's speech," E a r and Hearing 8(3 ) : 170-174. Dunn,  H.K. and White, S.D. (1940). measurements on c o n v e r s a t i o n a l speech," America 11:278-288.  "Statistical J.Acoust.Soc.of  82  F a n t , G. (1960). Acoustic Theory of Mouton, Hague. S e c o n d E d i t i o n 1970.  Speech  Production  F r e n c h , N.R. and S t e i n b e r g , J.C. (1947). " F a c t o r s g o v e r n i n g the i n t e l l i g i b i l i t y o f speech sounds," J . A c o u s t . S o c . o f A m e r i c a 19:90-119. Hirsh, I . J . , Reynolds, E.G. and Joseph, "Intelligibility of different speech J . A c o u s t . S o c . o f A m e r i c a 26:530-538. H o u t g a s t , T. a n d S t e e n e k e n , H.J.M. speech transmission channels s i g n a l s , " A c u s t i c a 25:355-367.  M. (1954). materials,"  (1971). "Evaluation of by using artificial  H o u t g a s t , T. a n d S t e e n e k e n , H.J.M. ( 1 9 7 3 ) . "The m o d u l a t i o n transfer function i n room a c o u s t i c s as a p r e d i c t o r o f s p e e c h i n t e l l i g i b i l i t y , " A c u s t i c a 28:66-73. H o u t g a s t , T. a n d S t e e n e k e n , H.J.M. ( 1 9 8 5 ) . "A r e v i e w o f t h e MTF concept i n room acoustics and i t s use f o r estimating speech intelligibility in auditoria," J . A c o u s t . S o c . o f A m e r i c a 77:1069-1077. H o u t g a s t , T. a n d S t e e n e k e n , H.M.J. ( 1 9 8 4 ) . language evaluation of the RASTI estimating speech intelligibility in A c u s t i c a 54:185-199. Houtgast, T., S t e e n e k e n , H.J.M. and " P r e d i c t i n g speech intelligibility modulation transfer function a c o u s t i c s , " A c u s t i c a 46:60-72.  "A multimethod f o r auditoria,"  Plomp, R. (1980). i n rooms f r o m t h e I. General room  H u d g i n s , C.V. (1960). "The d e v e l o p m e n t of communication s k i l l s among p r o f o u n d l y d e a f c h i l d r e n i n an a u d i t o r y training programme," I n I.R. Ewing (Ed.), Modern E d u c a t i o n a l Treatment o f Deafness. Manchester (Eng.): Manchester U n i v e r s i t y P r e s s . H u d g i n s , C.V., a n d Numbers, M. ( 1 9 4 2 ) . "An i n v e s t i g a t i o n o f the intelligibility o f t h e speech of the deaf," G e n e t . P s y c h o l o g . Monogr. 25:289-392. Humes, L . E . , D i r k s , D.D., Bell, T.S., A h l s t r o m , C. a n d K i n c a i d , G.E. (1986). " A p p l i c a t i o n o f the A r t i c u l a t i o n Index and t h e Speech Transmission Index to the r e c o g n i t i o n o f s p e e c h by n o r m a l - h e a r i n g and h e a r i n g impaired l i s t e n e r s , " J . o f Speech and H e a r i n g Res. 29:447-462. John,  J . E . J . , a n d Howarth, J.N., ( 1 9 6 5 ) . "The time distortions on t h e intelligibility c h i l d r e n ' s s p e e c h , " L a n g . S p e e c h 8:127-134.  effect of of deaf  83 Kamm, C.A., D i r k s , D.D. and B e l l , T.S. ( 1 9 8 5 ) . "Speech r e c o g n i t i o n and the a r t i c u l a t i o n i n d e x f o r n o r m a l and hearing-impaired listeners," J.Acoust.Soc.of America 77:281-288. K l e i n , W. (1971). " A r t i c u l a t i o n l o s s of consonants as a basis for the design and judgment of sOund r e i n f o r c e m e n t s y s t e m s , " J . A u d i o E n g . s o c . 19:11:920. K o n d r a s k e , G.V. (1985). " Q u a n t i t a t i v e assessment function," Proc. 7 t h . Annual Conference of Medical Biological Society. Chicago, S e p t e m b e r 27-30, 1985. Volume 2:671-674.  o f speech the IEEE Illinois.  K r y t e r , K.D. ( 1 9 6 2 a ) . "Methods f o r the c a l c u l a t i o n of the a r t i c u l a t i o n index," J.Acoust.Soc.of 34:1689-1697.  and u s e America  Kryter, K.D. (1962b). "Validation of the articulation i n d e x , " J . A c o u s t . S o c . o f A m e r i c a 34:1698-1702. K r y t e r , K.D. and B a l l , J.H. ( 1 9 6 4 ) . "SCIM - A meter f o r measuring the performance of speech communication systems," Decision Science Laboratory. Electronic systems D i v i s i o n . A i r F o r c e Systems Command. R e p o r t ESD-TDR-64-674. L i c k l i d e r , J.C.R., B i s b e r g , A. and S c h w a r z l a n d e r , H. ( 1 9 5 9 ) . "An e l e c t r o n i c d e v i c e t o measure t h e i n t e l l i g i b i l i t y o f speech," Proceedings of the N a t i o n a l Electronics C o n f e r e n c e 15:329. L o c h n e r , J . and B u r g e r , J. (1964). "The influence of r e f l e c t i o n s on auditorium acoustics," J.Sound Vibr. 4,426-454. Lundin, F . J . (1982). "The i n f l u e n c e o f room r e v e r b e r a t i o n on s p e e c h - an a c o u s t i c a l s t u d y o f s p e e c h i n a room," Speech T r a n s m i s s i o n L a b o r a t o r y , Q u a r t e r l y R e p o r t . STLQPSR 4/1982. Lundin, F . J . (1986). "A s t u d y o f speech intelligibility over a p u b l i c address system," Speech Transmission L a b o r a t o r y , Q u a r t e r l y R e p o r t . STL-QPSR 1/1986. M e t z , D.E., Sama R., V.J., Schiavetti, N., S i t l e r , R. and W h i t e h e a d , R.L. ( 1 9 8 5 ) . " A c o u s t i c d i m e n s i o n s o f h e a r i n g impaired speaker's i n t e l l i g i b i l i t y , " J . o f S p e e c h and H e a r i n g R e s . 28:345-355. M i l l e r , G.A. and Nicely, P.E. (1955). "An p e r c e p t u a l c o n f u s i o n s among some English J . A c o u s t . S o c . o f A m e r i c a 27:338-352.  analysis of consonants,"  84  Monsen, R.B. (1978). "Toward measuring how well deaf c h i l d r e n speak," J . o f S p e e c h a n d H e a r i n g R e s . 21:197219. M o r s e , P.M. and Ingard, M c G r a w - H i l l . N.Y.  U.K.  (1968). T h e o r e t i c a l A c o u s t i c s  Newman, P.W ( 1 9 7 9 ) . "Appraisal of a r t i c u l a t i o n , " Part IIi n Darley, F.L. (Editor) Evaluation of Appraisal T e c h n i q u e s i n S p e e c h a n d Language P a t h o l o g y . Don M i l l s , O n t a r i o : A d d i s o n - W e s l e y P u b l i s h i n g Company. P a v l o v i c , C.V. ( 1 9 8 7 ) . " D e r i v a t i o n o f p r i m a r y p a r a m e t e r s a n d procedures for use in speech intelligibility p r e d i c t i o n s , " J . A c o u s t . S o c . o f A m e r i c a 82:413-422. P a v l o v i c , C.V. a n d S t u d e b a k e r , G.A. (1984) "An e v a l u a t i o n o f some a s s u m p t i o n s underlying the a r t i c u l a t i o n index," J . A c o u s t . S o c . o f A m e r i c a 75:1606-1612. P e u t z , V.M.A. ( 1 9 7 1 ) . " A r t i c u l a t i o n l o s s o f consonants as a criterion for speech transmission in a room," J.Audio.Eng. 19(11):915. Picheny, M.A., Durlach, N . I . a n d B r a i d a , L.D. (1985). "Speaking clearly f o r the hard of hearing I: Intelligibility differences between clear and c o n v e r s a t i o n a l speech," J . o f Speech and H e a r i n g Res. 28:96-103. Picheny, M.A., Durlach, N . I . a n d B r a i d a , L.D. (1986). "Speaking c l e a r l y f o r the hard of hearing I I : Acoustic c h a r a c t e r i s t i c s o f c l e a r and c o n v e r s a t i o n a l speech," J . o f S p e e c h a n d H e a r i n g R e s . 29:434-446. Plomp, R., Houtgast, T. a n d S t e e n e k e n , H.J.M. (1980). " P r e d i c t i n g speech i n t e l l i g i b i l i t y i n rooms f r o m t h e modulation t r a n s f e r f u n c t i o n I I . M i r r o r image c o m p u t e r model a p p l i e d to rectangular rooms," J.Acoust.Soc.of A m e r i c a 46:73. R e t t i n g e r , M.A. (1968). A c o u s t i c s : Room Control New Y o r k : C h e m i c a l P u b l i s h i n g S c h r o e d e r , M.R. (1981). A c u s t i c a 49:179.  Design and Noise Co., I n c .  "Modulation t r a n s f e r  functions,"  S t e e n e k e n , H.J.M. and A g t e r h u i s , E. (1978). " D e s c r i p t i o n o f STIDAS I I - c (Speech T r a n s m i s s i o n Index Device using A r t i f i c i a l Signals) Part I , " Report 1978-19,IZF-TNO, S o e s t e r b e r g , The N e t h e r l a n d s .  85  S t e e n e k e n , H.J.M. and H o u t g a s t , T. (1980). "A physical method for measuring speech-transmission quality," J . A c o u s t . S o c . o f A m e r i c a 67:318-326. S t e e n e k e n , H.J.M. a n d H o u t g a s t , T. ( 1 9 8 3 ) . "The t e m p o r a l envelope spectrum o f speech and i t s s i g n i f i c a n c e i n room acoustics," In Proceedings of the Eleventh I n t e r n a t i o n a l C o n g r e s s on A c o u s t i c s . 7:85-88. S t e e n e k e n , H.J.M. and H o u t g a s t , T. ( 1 9 8 5 ) . "RASTI: A t o o l f o r e v a l u a t i n g a u d i t o r i a , " B r u e l & K.jaer T e c h . Rev. 3. van  R e i t s c h o t e , H.F., H o u t g a s t , T. a n d S t e e n e k e n , H.J.M. (1981). " P r e d i c t i n g speech i n t e l l i g i b i l i t y i n rooms f r o m t h e m o d u l a t i o n t r a n s f e r f u n c t i o n IV: A r a y - t r a c i n g c o m p u t e r m o d e l , " A c u s t i c a 49:245-252.  van R e i t s c h o t e , H.F., H o u t g a s t , T. a n d S t e e n e k e n , H.J.M. (1984). " P r e d i c t i n g speech i n t e l l i g i b i l i t y i n rooms f r o m t h e m o d u l a t i o n t r a n s f e r f u n c t i o n V: The m e r i t s o f a ray-tracing model v e r s u s general room acoustics," A c u s t i c a 53:72-78. Voelker, CH. (1938) "An experimental study of the comparative rate of u t t e r a n c e o f deaf and normal h e a r i n g s p e a k e r s , " Am.Ann.Deaf 83:274-284. W i l l i a m s , C.E. a n d H e c k e r , M.H.L. ( 1 9 6 8 ) . " R e l a t i o n between i n t e l l i g i b i l i t y scores f o r types o f speech d i s t o r t i o n . J . A c o u s t . S o c . o f A m e r i c a 44:1002.  86  APPENDIX A I N T E L L I G I B I L I T Y TESTS BASED ON  A.  TESTS CITED BY  BLACK  LISTENER  JUDGMENTS  (1957)  B e l l Telephone's Tests reference: Fletcher, H. "Articulation testing .1^8:806-854.  and Steinberg, methods," Bell  J.C. (1929). Syst. Tech.  H a r v a r d ' s P h o n e t i c a l l y B a l a n c e d (PB) T e s t s r e f e r e n c e s : a) Egan, J.P. (1944). "Articulation testing methods II," Psycho-Acoustic Laboratory, Harvard U n i v e r s i t y , Nov. OSRD R e p t . no. 3820. b) Egan, J.P. (1948). " A r t i c u l a t i o n t e s t i n g methods," Laryngoscope 58:955-991. V o i c e Communication L a b o r a t o r y ' s T e s t reference: Haagen, CH. (1944). "Intelligibility measurement: T e c h n i q u e s and procedures used by the Voice Communication Laboratory," Psychological C o r p o r a t i o n . , New Y o r k , OSRD R e p t . no. 3748. B.  TESTS REVIEWED IN NEWMAN'S  D r u m w r i g h t , A.F. Examination Foundation,  (1971). (DASE) Inc.  (1979) CHAPTER  The D e n v e r A r t i c u l a t i o n S c r e e n i n g Ladoca Project and Publishing  F i s h e r , H.A. and Logemann, J.A. ( 1 9 7 1 ) . The F i s h e r - L o g e m a n n Test of Articulation Competence Houghton Mifflin Company. F u d a l a , J.B. (1963). A r i z o n a A r t i c u l a t i o n P r o f i c i e n c y S c a l e Western P s y c h o l o g i c a l Services, (later e d i t i o n s 1970, 1974). Goldman, R. and F r i s t o e , M. (1969). Goldman-Fristoe o f A r t i c u l a t i o n (GFTA) American Guidance S e r v i c e , ( l a t e r e d i t i o n p u b l i s h e d 1972). H e j n a , R.F. (1968). Materials.  Developmental A r t i c u l a t i o n  Test  Test Inc.  Speech  I r w i n , O.C (1972) " I n t e g r a t e d a r t i c u l a t i o n t e s t , " I n O r v i s C. I r w i n , Communication V a r i a b l e s of Cerebral P a l s i e d and Mentally Retarded Children Springfield, 111: C h a r l e s C. Thomas.  M c D o n a l d , E.T. ( 1 9 6 4 ) . A Deep T e s t o f A r t i c u l a t i o n . P i c t u r e Form S t a n w i x House, I n c . ( A l s o a v a i l a b l e f r o m t h e same p u b l i s h e r are the Sentence and S c r e e n i n g Forms o f t h e same t e s t ) . Mecham, J . L . J . and J o n e s , J.D. (1970). S c r e e n i n g Speech Articulation Test (SSAT) Communication Research A s s o c i a t e s , Inc.. P e n d e r g a s t , K., D i c k e y , S., S e l m a r , J.W., and S o d e r , A.L. (1969). Photo A r t i c u l a t i o n Test (PAT) Interstate P r i n t e r s and P u b l i s h e r s , I n c . . T e m p l i n , M.C. and D a r l e y , F.L. (1969). The T e m p l i n - D a r l e y Tests of Articulation U n i v e r s i t y of Iowa B u r e a u of E d u c a t i o n a l R e s e a r c h and S e r v i c e , Sound e d i t i o n . T o r o n t o , A.S. (1977). Southwestern Spanish T e s t (SSAT) Academic T e s t s , Inc..  Articulation  Van  Riper, C. and Erickson, R.L. ( 1973) . Predictive S c r e e n i n g T e s t o f A r t i c u l a t i o n (PSTA) Western M i c h i g a n University, Continuing Education O f f i c e , Third E d i t i o n .  C.  TESTS CITED BY  WILLIAMS AND  HECKER  (1968)  H a r v a r d PB-Word I n t e l l i g i b i l i t y T e s t and Harvard Sentence Tests reference: Egan, J.P. (1948) "Articulation methods," L a r y n g o s c o p e 58:955-991. F a i r b a n k s Rhyme T e s t reference: Fairbanks, d i f f e r e n t i a t i o n : The 30:596-600.  G. (1958) Rhyme T e s t , "  "Test of J.Acoust.  testing  phonemic Soc.Amer.  M o d i f i e d Rhyme T e s t reference: House, A.S., W i l l i a m s , C.E., H e c k e r , M.H.L and K r y t e r , K.D. (1965) "Articulation t e s t i n g methods: Consonantal differentiation with a closed-response s e t , " J . A c o u s t . Soc. Amer. 37:158-166.  APPENDIX THE ENGLISH AND  B  FRENCH SENTENCES  SELECTED  ENGLISH 1. 2. 3.  P a t t y put f i v e pennies i n her purse. Shallow seas a r e not shark i n f e s t e d . G r e e n g r a p e v i n e s grow i n c o u n t r y g a r d e n s .  FRENCH 1. 2. 3.  I I n'y a j a m a i s de fumee s a n s f e u . I I j o u e du trombone t o u s l e s l u n d i s . Un g r a n d c o c a - c o l a s a n s g l a c o n s .  89 APPENDIX C LISTENING TEST INSTRUCTIONS AND ANSWER SHEET PART 1 - LISTENING TEST ANSWER SHEET Subject  ID: L  In this experiment, you w i l l be h e a r i n g a number o f s e n t e n c e s spoken by s i x E n g l i s h speakers a n d one F r e n c h speaker. Sometimes the sentences are mumbled ( u n d e r a r t i c u l a t e d ) , sometimes t h e y a r e a r t i c u l a t e d n o r m a l l y , and s o m e t i m e s they a r e o v e r a r t i c u l a t e d . Your t a s k i s to a s s i g n a number on a s e v e n point scale f o r each s e n t e n c e i n d i c a t i n g how i t sounds t o you. F o r example, i f y o u were f a i r l y c e r t a i n t h a t a s e n t e n c e was n o r m a l l y a r t i c u l a t e d , y o u w o u l d c i r c l e t h e number 4 a s shown b e l o w : mumbled  141.  1  normal  2  3  whereas i f y o u t h o u g h t t h a t c i r c l e 1 a s shown b e l o w : mumbled 142.  1  4 item  overarticulated  5  3  4  7,  142 was mumbled, y o u m i g h t  normal 2  6  overarticulated 5  6  7  The s c a l e r e p r e s e n t s a c o n t i n u u m between mumbled (number 1) and o v e r a r t i c u l a t e d (number 7 ) . You may c h o o s e a n y number w h i c h y o u t h i n k i s a p p r o p r i a t e t o what y o u h e a r . Please t r y to attend o n l y t o whether t h e s e n t e n c e sounds mumbled, normally articulated, o r o v e r a r t i c u l a t e d , and ignore other v a r i a b l e s such as d i f f e r e n c e s i n r e c o r d i n g o r voice quality, speed o f a r t i c u l a t i o n , language spoken, o r sentence content. P l e a s e b e g i n by l i s t e n i n g t o t h e f i r s t t e n s e n t e n c e s w i t h o u t marking t h e paper, and then s t o p t h e tape and ask q u e s t i o n s i f y o u need c l a r i f i c a t i o n about any p a r t of the task. Following this, t h e tape w i l l be rewound t o t h e b e g i n n i n g and y o u w i l l be a s k e d t o l i s t e n t o t h e e n t i r e t a p e a n d mark y o u r answer s h e e t , w i t h o u t r e w i n d i n g o r s t o p p i n g t h e t a p e i f possible.  Subject ID: L mumbled 1.  1  2  2.  1  2  3.  1  2  4.  1  2  5.  1  2  6.  1  2  7.  1  2  8.  1  2  129.  1  130.  3  normal 4  5  overarticulated 6 7  3  4  5  6  7  3  4  5  6  7  3  4  5  6  7  3  4  5  6  7  3  4  5  6  7  3  4  5  6  7  3  ( ...i e .tc• • • )  2  3  4  5  6  7  1  2  3  4  5  6  7  131.  1  2  3  4  5  6  7  132.  1  2  3  4  5  6  7  133.  1  2  3  4  5  6  7  134.  1  2  3  4  5  6  7  135.  1  2  3  4  5  6  7  136.  1  2  3  4  5  6  7  137.  1  2  3  4  5  6  7  138.  1  2  3  4  5  6  7  139.  1  2  3  4  5  6  7  140.  1 mumbled  2  3  4 normal  5 overarticulated  APPENDIX LISTING  D  OF THE FOCAL-12 PROGRAM FOR THE MODULATION INDEX  CALCULATING  PART A 1 .01 1 .03 1.05 1 .07 1.08 1 .10 K=l 1.20  C PROGRAM DHPP12A E O C L O,F0,I,INPUT,l L O,F0,I,OUTPUT,0 A "FIRST SAMPLE",SF,!,"LAST D 2;G  SAMPLE",SL, !;S  I=SF+2  4.1  2. 10 S A=F0(I)-F0(I-1);S B(0)=F0(I)-F0(I+1) 2.20 S B(I)+F0(I+l)-F0(I+2);S B(2)=F0(I+2)-F0(I+3);S B(3) = F0(1+3)-F0(1+4);S B(4)=F0(I+4)-F0(I+5) 4. 10 4.20 4 . 30  I (A)4.2,4.3;I (B(0))5.4,5.2,5.1 I (B(0))5.6,5.3,5.4 I (B(0))5.2,5.5,5.2  5.09 C IT IS A PEAK 5.10 S F l ( K ) = F 0 ( I ) ; S F1(K+1)=I;S K=K+2;G 6.1;C STORING AMPLITUDE AND LOCATION OF PROUGHS 5.19 C SHOULD NOT HAPPEN 5.20 G 7.1 5.29 C LOOK ONE, TWO, OR THREE AHEAD 5 . 30 I ( A * B ( 1 ) ) 5 . 8 , 5 . 3 1 ; I (A)5.34,7.1,5.34 5.31 I ( A * B ( 2 ) ) 5 . 8 5 , 5 . 3 2 ; L I (A)5.35,7.1,5.35 5.32 I (A*B(3))5.9,5.33;I (A)5.37,7.1,5.37 5.33 I ( A * B ( 4 ) ) 5 . 9 5 , 7 . 1 ; I (A)5.38,7.1,5.38 I=I+1;S K+K+2;G 6 5.34 S F1(K)=F0(I);S Fl(K+l)=1+0.5;S 5.35 s F 1 ( K ) = F 0 ( I ) ; S F1(K+1)=I+1.0;S I=I+2;S K=K+2;G 6 5.37 s F l ( K ) = F 0 ( I ) ; S F1(K+1)=I+1.5;S I=I+3;S K=K+2;G 6 5.38 s F 1 ( K ) = F 0 ( I ) ; S F l ( K + l ) = I + 2 . 0 ; S I=I+4;S K=K+2;G 6 5.39 C CONTINUE 5.40 G 6.1 5.49 C SHOULD NOT HAPPEN 5.50 G 7.1 5.59 C IT IS A TROUGH 5.60 S F l ( K ) = F 0 ( I ) ; S F1(K+1)=I;S K=K+2;G 6.1 5.80 S 1=1+1;G 6.1 5.85 S I=I+2;G 6.1 5.90 S I=I+3;G 6.1 5.95 S 1=1+4;G 6.1 6.10 S I=I+l;I (I-SL+5)1.2,1.2 6.20 T %2.01,!!,1,!,K,!!!."TYPE RETURN TO CONTINUE";A 6 . 30 S F 1 ( 0 ) = ( K - 1 ) / 2 ; C STORES IN F 1 ( 0 ) THE NUMBER OF "PROUGHS"  !  6.40 6.50  L C,F1 L O.DHPP12B.0  PART B PROGRAM  DHPP12B  1.01 1.02 1.03 1.05 1.10 1.20 1.30 1.40 1.50 1.60 1.90  C E 0 L S F D S S T Q  2.01 2.10 2.20 2.30  C I I S  DETERMINING MAX AND MIN ( F 1 ( I ) - M A X ) 2 . 2 , 2 . 2 ; S MAX=F1(I) ( M I N - F l ( I ) ) 2 . 3 , 2 . 3 ; S MIN=F1(I) AV=(MA+MI)/2;R  3.01 3.10 3.20 3.30 3.40 3.50 3.90  c I s I s s T  DETERMINING THE LOCATION OF THE FIRST AND LAST PEAKS ( F 1 ( I ) - F 1 ( 3 ) ) 3 . 2 , 3 . 9 ; S IS=1;G 3.3 IS = 3 ( F l ( K * 2 - l ) - F l ( K * 2 - 3 ) ) 3 . 4 , 3 . 9 ; S IL=K*2-1;G 3.5 IL=K*2-3 N P = ( I L - I S ) / 4 + l ; S NT=NP-l;R "TROUBLE',!;Q  C 0,F1,F,OUTPUT,0 MAX=0;S MIN=1000;S K=F1(0) I+1,2,K*2;D 2 3 P R = A V / F 1 ( I S ) ; F I=IS+2,4,IL-2;S MI=FEXP((l/NP)*FLOG(PR)) %7.06,"MI = ",MI,!!  PR=PR*F1(I)/FI(1+2)  

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:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0097698/manifest

Comment

Related Items