Synthesized Speech

Liner notes:

A MACHINE THAT TALKS:

There are many possible kind of synthesizers or "talking" machines. To save the expense and time of building, testing, and modifying them, John L. Kelly, Jr. and Louis J. Gertsman of the Visual and Acoustics Research Department at Bell Laboratories use a high-speed, general purpose computer to simulate them.

The computer is instructed to accept cards, to "operate" on this information similarly to the way an actual talking machine does, and to produce an "output" analogous to the output of the talking machine.

By changing the computer program it is comparatively easy to modify the characteristics of the talking machine.

The particular machine which was simulated in the computer to produce the speech on the recording is known technically as a "tandem resonant synthesizer." Usually this type of machine is operated by continuously feeding into it as a set of nine signals corresponding to voice pitch, voice loudness, tongue position, and other speech variables.

When every instant of sound is specified, the machine produces sounds that are amazingly human like speech.

A PHONETIC INPUT:

Doctors Kelly and Gerstman have contributed a significant advance in the art of speech synthesis by devising a computer program which permits them to feed into the computer, on punched cards, the names of speech sounds.

Since the standard phonetic symbols representing speech sounds are not included on the keyboard of an ordinary card-punching machine, Kelly and Gerstman devised a new phonetic code using the letters of the alphabet.

At present it consists of 22 consonants and 12 vowels:

CONSONANTS: P - B - T - D - K - G - M - N - NG (as in sing) - F - V - S - Z - SH (as in she) - ZH (as in azure) - H - W - R - L - Y - TH (as in thin) - DH (as in then)

VOWELS: EE (as in bee) - I (as in ill) - AY (as in rate) - E (as in end) - AE (as in add) - AH (as in ah) - AW (as in jaw) - (as in go) - OO (as in foot) - UU (as in food) - UH (as in up) - ER (as in her)

Each speech sound is specified on a separate punched card.

When a sequence of cards is fed into the computer, it "operates' on the information - following the rules set up in the second part of its program - to produce the nine control signals that activate the talking machine program. For example, if the sequence of cards, H,EE,S,AW,DH,UH,K,AE, T is put into the computer, the talking machine will say "He Saw The Cat" in measured monotone voice.

To obtain natural intonation and phrasing it is necessary to specify on each card (in addition to the speech sound) both the pitch of the sound and timing information.

A SPEECH LIKE OUTPUT

The "Speech" of the simulated talk machine comes out of the computer in the form of tiny magnetized spots on half-inch magnetic tape. This tape is fed to another machine which converts the digital information to a variable magnetic sound track suitable for playing on an ordinary tape recorder playback.

ON THIS RECORDING

The samples of speech on this recording illustrate the present state-of-the-art of speech synthesis.