Patent No. 3647970 Method and system for simplifying speech waveforms

 

Patent No. 3647970  Method and system for simplifying speech waveforms (Flanagan, Mar 7, 1972)

Abstract

A speech waveform is converted to a constant amplitude square wave in which the transitions between the amplitude extremes are spaced so as to carry the speech information. The system includes a pair of tuned amplifier circuits which act as high-pass filters having a 6 decibel per octave slope from 0 to 15,000 cycles followed by two stages, each comprised of an amplifier and clipper circuit, for converting the filtered waveform to a square wave. A radio transmitter and receiver having a plurality of separate channels within a conventional single side band transmitter bandwidth and a system for transmitting secure speech information are also disclosed.

Notes:

BACKGROUND OF INVENTION

This invention relates generally to electronic processing of speech, and more particularly relates to a method and system for simplifying the speech waveform to facilitate transmission of the speech through various media without materially degrading intelligibility.

In the process of producing human speech, the voice box creates a series of sound pulses which reverberate within and are shaped by the upper throat and mouth cavity. The frequency of the pulses produced by the voice box primarily determines the frequency or pitch of the sound, while the shape of the mouth cavity reverberates and shapes the sound pulses to produce the speech information. The resulting speech waveform is very complex and highly redundant. If such a waveform is passed through a band-pass filter having a bandwidth significantly less than 3,000 cycles per second, the speech becomes unintelligible. Thus, even the simplest voice communication channels require a substantial bandwidth. Heretofore it has been commonly believed that the speech information was contained in the amplitude as well as the frequency modulation of the speech waveform. When voice sounds are induced in a body of water or the earth, the many reverberations caused by the various velocity discontinuities make speech unintelligible over relatively short transmission lengths. Also, the complex speech waveform has made encoding or scrambling for secure transmissions, either by electromagnetic, electrical, or pressure waves, so impractical as to be very seldom used.

SUMMARY OF INVENTION CLAIMED

This invention is concerned with a method and system for simplifying a complex speech waveform so that it can be used for a multitude of applications. The simplified speech waveform may be passed through a narrow band-pass filter, thus permitting a greater number of communication channels within a given frequency band. The simplified speech waveform can be transmitted directly through the earth or water as a pressure wave and understood, either directly from the medium, or after simple amplification. The simplified waveform can be easily encoded by scrambling to provide secure voice communications. The simplified waveform may be used to operate machinery, produces more efficient public address systems and transmitters with greater range peak power for a given average power, and thus longer ranges.

In accordance with the present invention, the speech waveform is converted to a signal having substantially constant upper and lower levels with abrupt transitions from one level to the other, the abrupt transitions being in time correspondence to amplitude changes in the speech waveform that exceed a predetermined rate of change. This is accomplished by a system including a high-pass filter and means for converting the filtered waveform to a constant amplitude, substantially square wave.

More specifically, optimum results have been achieved by using a filter having a 12 decibel per octave slope from 0 to 15,000 cycles per second. In one specific embodiment, this filter is formed by a pair of tuned amplifier circuits each having a 6 decibel per octave slope within the frequency range of interest. In this embodiment, the speech waveform is preferably combined with a high frequency noise masking signal of lower amplitude prior to processing.

In accordance with another specific aspect of the invention, means for converting the filtered signal to a square wave comprises at least one amplifier followed by a clipper circuit.

The invention also contemplates a voice communication system having a plurality of separate channels within a bandwidth normally allotted for a single frequency, for example four channels within a bandwidth of 1,500 cycles per second. In this system the processed speech is selectively passed through one of a plurality of narrow band-pass filters to a transmitter. The receiver has similar narrow band-pass filters so as to be selectively sensitive to transmissions in that pass band.

In accordance with another specific aspect of the invention, each transition of the square wave is converted to a pulse of predetermined amplitude and width, which is then converted into a plurality of pulses with predetermined time spacing. These pulses are then transmitted to a receiver where the plurality of spaced pulses are recombined as one pulse. The square wave is then reproduced from the recombined pulses.

Although preferred embodiments of the invention have been described in detail, it is to be understood that various changes, substitutions and alterations can be made therein without departing from the spirit and scope of the invention as defined by the appended claims.