Effects of Distortion Upon the Recognition of Speech Sounds
John C. Steinberg · The Journal of the Acoustical Society of America · 1929
Physical studies of speech waves in conjunction with articulation tests have shown that in normal speech, the fundamental sounds are distributed in intensity and frequency. For example, the vowels are some 20 to 30 db greater in intensity than the fricative consonants, and require the range of frequencies, from a recognition standpoint, from 500 to 2500 cycles as compared with the range from 1500 to 8000 cycles for the fricatives. Some sounds, such as the vowels, tend to approach a “steady state” condition; others, such as the stop consonants, are of relatively short duration and almost “transient” in character. In both cases, however, recognition appears to depend considerably upon the initial period of growth and the final period of decay. With these facts in mind, the primary effects of distortion may be readily understood. Distortion which tends to increase the intensity spread between sounds, or which effectively lowers the received speech level, decreases the articulation of the less intense sounds. Distortion which suppresses a portion of the frequency range, decreases the articulation of the sounds that are characteristic of that range. Systems which overload for the intense sounds tend to decrease the intensity spread, and may therefore increase the articulation of the faint sounds while decreasing the articulation of the intense sounds. Reverberation affects the various sounds differently, the greatest impairment occurring for the nasal and stop consonants. In no case did an increase in reverberation time cause an increase in articulation.