SPEECH RECOGNITION WITH PRIMARILY TEMPORAL CUES

Nearly perfect speech recognition was observed under conditions of gre atly reduced spectral information. Temporal envelopes of speech were e xtracted from broad frequency bands and were used to modulate noises o f the same bandwidths. This manipulation preserved temporal envelope c ues in each band but restricted the listener to severely degraded info rmation on the distribution of spectral energy. The identification of consonants, vowels, and words in simple sentences improved markedly as the number of bands increased; high speech recognition performance wa s obtained with only three bands of modulated noise. Thus, the present ation of a dynamic temporal pattern in only a few broad spectral regio ns is sufficient for the recognition of speech.