UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Computer simulation, development and evaluation of a high speed spelled speech code Suen, China Yee


A high speed spelled speech code has been developed for a reading machine for the blind. Keeping within the constraints of high reading speed and high intelligibility, a main contribution of the work has been to minimize the memory size and thus the cost of the digital spelled speech reading machine. In order to reduce the amount of memory required to generate letter sounds of this code, redundant phonemes were eliminated and a selected set of 18 basic phonemes was extracted by a segmentation program. Letter sounds were then synthesized by concatenation of these basic phonemes. Also, vowels and vowel-like sounds have quasi-periodic waveforms. These sounds were reproduced satisfactorily by repeating over and over again a pitch period extracted from the original waveforms. Another reduction of digital memory storage was accomplished by .providing each individual phoneme with a minimum number of bits per sample. The segmentation program developed runs on a PDP-9 digital computer. This program has the functions of acquisition, graphic display, data print-out, auditory presentation, manipulation and extraction of speech samples. Graphic display of the amplitude-time waveforms of various segments of a speech sample provided an accurate and efficient method of extracting the basic phonemes. Six vowels extracted in this way were experimented in a discrimination test. It was found that even when these vowels were only 10 ms. in duration, the subjects could learn to discriminate them. The PDP-9 computer was also used to synthesize the letter sounds and to simulate a spelled speech machine. Experiment with three blind subjects indicated that they could read spelled sentences between 60 and 70 words per minute with high intelligibility after only one hour of contact with this spoiled speech code. A difference coding scheme was used to reduce further the amount of digital memory required to store the basic phonemes. An attempt was also made to find out whether memory storage could be reduced by lowering the sampling rate. This was studied by reducing the bandwidth of the letter sounds in a subjective test using 16 blind students. Also investigated in this experiment were the intelligibility of the letter sounds and the effects of presentation speed and pause between words. Experimental results confirmed that most blind subjects could learn to recognize all 26 synthesized letter sounds after a short period of training, and they could read spelled sentences between 65 and 75 words per minute with an intelligibility score of about 85% correct. Bandwidth reduction reduced the pleasantness and clarity of the letter sounds. It was concluded that for the reduction of memory storage, the difference coding scheme was preferred and the original bandwidth of 6 kHz. should be retained.

Item Media

Item Citations and Data


For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.