Pattern Playback: Background Information

The use of the Pattern Playback is based around an experimental procedure that involves two steps, first the representation of speech in terms of a sound spectrogram, and second, the reconversion of these visual patterns -- after appropriate simplifications and modifications -- back into sound for evaluation by ear. The operating principle of the Pattern Playback employs about the simplest possible way of reinserting the frequencies specified by the spectrogram. A tone wheel supplies the first 50 harmonics of a fundamental frequency of 120 Hertz (Hz), thus giving a frequency range of 0 to 6000 Hz. Because of this fixed fundamental frequency, the device will “speak” only in a monotone.

The spectrogram from which the sound is reconstituted can be a photographic copy of an original spectrogram, in which case the appropriate frequencies are transmitted by the film to the light collector and photo-cell. For synthetic spectrograms, on the other hand, it is most convenient to paint only what one wishes to hear. Hence, flat white paint on clear cellulose acetate is used, and the useful light is that which is scattered back into the light collector by diffuse reflection from the painted areas. Convenience of operation was a most important consideration in the design of this equipment.

The Pattern Playback created signals perceivable as speech sounds by reproducing the critical acoustic information in the speech signal. Such critical information consists of the formant patterns (main bands of energy) resulting from the resonance characteristics of a dynamically changing vocal tract, along with the other noises, hisses, pops, clicks, etc., that can be seen in a frequency analysis of a signal, often done with a sound spectrograph (also known as a voiceprint).

<   Pattern Playback   >