AES E-Library

AES E-Library

Fundamentals of Speech Synthesis

Document Thumbnail

From the communication engineer's point of view, there are two fundamental principles of human speech synthesis. The first is that speech is a message, which, in its initial expression from the body, is represented by a group of muscular vibrations-a multiplicity of telegraph signals functionally similar to the muscular vibrations of the finger in keying the simplest type of telegraph on-off signals. The second principle is that these -multiple telegraphed signals- are made audible to the ear by modulating a set of carrier-frequency components in the audible part of the spectrum. Such components range from an inflected, buzz-like tone for voiced sounds to a hiss-like noise for unvoiced sounds, including whispers. The Vocoder was an early development of apparatus exploiting both of these basic characteristics of speech-the telegraph nature in the analyzer and the carrier nature in the synthesizer. If advantage is taken of the telegraph nature of speech, a large reduction can be effected in the frequency band required for transmission. Where long, expensive lines are concerned, economic benefit to the telephone user may accrue from such a reduction in bandwidth. The ultimate in thus telegraphizing speech for transmission would be obtained in a system able to recognize phonetic elements and then transmit them by narrowband telegraph methods. Rudimentary experiments along these lines have been performed, using -Audrey,- the automatic digit recognizer, to recognize the individual phonetic units of the sending end, and using a Vocoder at the receiving end for the purpose of synthesizing somewhat standardized speech from the telegraph-like currents transmitted from the sending end. Other basic speech analyzer-synthesizer devices experimented with at the Bell Telephone Laboratories over the years are briefly reviewed; these devices include the Vocoder, the sound spectrograph, the visible-speech translator and synthesizer, the electrical vocal tract, and the improved Vocoder.

JAES Volume 3 Issue 4 pp. 170-185; October 1955
Publication Date:

Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!

This paper costs $33 for non-members and is free for AES members and E-Library subscribers.

Learn more about the AES E-Library

E-Library Location:

Start a discussion about this paper!

AES - Audio Engineering Society