Search

Many new improvements over earlier models of the phonetic typewriter have resulted from developments which provide means for reducing the number of different displays obtained for the same syllable without unduly increasing the complexity of the machine. In terms of the ultimate machine which should provide means for handling 1000 to 2000 syllables, model III with a maximum capacity of about 100 syllables is still a rudimentary phonetic typewriter. However, the model III provides a complete system from the sounds of speech input to the typed‐page output which is of great value in future research in speech analysis. Specifically, the subjects considered in this paper are as follows: form of the typed presentation, construction, circuits, syllable memory, typing console, spelling memory, choice of vocabulary, procedure for setting up the memory, and codes obtained.

A public demonstration of a rudimentary speech processing system consisting of a speech analyzer, language translator, type decoder, and speech synthesizer was given at the Meeting of the Acoustical Society of America in New York, May 23–26, 1962. Each word spoken into the microphone was analyzed, converted to a syllable code, transmitted as a syllable code, synthesized to speech, or converted to the typed page either in the same language or other languages. The two language inputs were English and French, the synthesized speech outputs were English, French, and Spanish and the typed outputs were English, French, German, and Spanish.

A printout system has been developed which automatically records any one of over a trillion different quantized analyses of a spoken syllable. The main purpose is research toward improved methods of recognizing speech by machines. However, other applications for the digitized transmission of speech are also indicated. Results of tests made are analyzed and different approaches to the decoding problem are presented.

The electronic music synthesizer is a machine that produces music from a coded record. The coded record, is produced by a musician, musical engineer, or composer with a fundamental understanding of the composition of sound. The electronic music synthesizer provides means for the production of a tone with any frequency, intensity, growth, duration, decay, portamento, timbre, vibrato, and variation. If these properties of a tone are specified, the tone can be completely described. The advantage of the electronic music synthesizer is that it can produce new and radical tone complexes for musical satisfaction and gratification. The new system does not displace the artist and musician of today. It does not take the place of talent combined with work. The electronic music synthesizer provides the musician, musical engineer, and composer with a new musical tool with no inherent physical limitations.

The random probability system generates sequences of notes in a series which are not completely random nor completely ordered, but each note is selected in a random fashion with a probability which depends upon preceding notes. Combined with sound synthesizers, the machine provides additional instrumentation for the musical laboratory.

A speech analyzer and synthesizer with a capacity of 200 speech elements has been developed with the objective of evaluating the performance. The speech analyzer consisting of normalizers, syllable and syblet segmenters, sound recognizers, sequencers, and memories converts the speech elements into a code. The speech synthesizer converts the code to speech by reproducing prerecorded speech elements. The advantages of the voice‐communication system employing the analyzer and synthesizer are the high intelligibility and the low transmission rate of only 23 bit/sec.

A simplified version, but nevertheless complete system, of a phonetic typewriter which types in response to words spoken into a microphone has been described [H. F. Olson and H. Belar, J. Acoust. Soc. Am. 28, 1072–1081 (1956)]. One of the many extensions of the original system is a means for compensating for the effects of the speed of talking. Instead of transferring information into the spectral memory in fixed time increments, as was done in the original machine, the transformation of information is determined from a correlation with significant changes in the speech spectrum with respect to time. In the improved machine, coincidence detectors compare the current with the preceding speech spectrum. No information is fed into the spectral memory until a significant change has been detected in the speech spectrum. The system provides an automatic time compensator for variations in the speed of talking in a speech recognition machine.

The modern systems of communication have been applied to the electronic synthesis of music in the form of the electronic music synthesizer. The electronic music synthesizer is a machine which provides means for producing a tone with any frequency, intensity, growth, duration, decay, portamento, timbre, vibrato, and deviations. If these properties of a tone are specified, the tone can be produced by the synthesizer from a coded record. For implementing the electronic synthesis of music an improved and enlarged electronic music synthesizer Mark II has been developed and built. The new electronic music synthesizer and the use of the new synthesizer by professional musicians will be described.

A syllable synthesizer is required to complete a voice‐to‐voice communication system in which the sender speaks one syllable at a time into a microphone and in which transmission to the receiver takes place in the form of a syllable number. The merit of such a system is the extremely small communication‐channel capacity. Speech synthesized from prerecorded syllables has been proven highly intelligible. To determine the possibility of synthesizing speech from segments of spoken syllables, a 200‐syllable synthesizer was used as an investigative tool. Statistical studies were made in order to find the smallest recognizable particle of a spoken syllable that can be segmented by machine, recognized by machine, and from which speech can be synthesized by machine from prerecordings. This particle has been tentatively given the name “Syblet,” a contraction of syllablet. The design of the 200‐syllable synthesizer is described, the results of listening tests presented, and a demonstration of speech from prerecorded syblets is given.

The following speech‐processing systems are described and demonstrated: (1) phonetic typewriter—the conversion of speech to the typed page; (2) speech analysis, conversion to a code, transmission of the coded signal, and the synthesis of speech from the code; (3) application of language translation of speech to the typed page and to synthesized speech.