New Instrumentation for Making Spectrographic Pictures of Speech

Publisher Website
Google Scholar
Abstract
Communication by speech involves the production and perception of elements of an acoustic code. This code can be made visible in pictures which show certain spectralpatterns existing in the sound signals. A new instrumentation has been developed for making spectrographic pictures of large quantities of speech. The outputs of a bank of filters are repeatedly scanned at a high rate to produce spectral cross sections. Each cross section presents the existing spectrum as a continuous distribution of energy versus frequency. A unique process of interpolation of filter outputs maximizes the information recovered so that the frequencies of tones are accurately represented by peaks in the spectral cross sections. These cross sections may be recorded on film to provide a complete and detailed time history of signal spectra. Other picture formats include spectrograms showing amplitude by density modulation. Strip‐film recording and operation in real time combine to provide a very great analyzing capability. This paper reviews the basic analyzing functions, describes some of the kinds of flexibility of analysis, and presents reproductions of films made by the analyzer.

This publication has 0 references indexed in Scilit: