Relative contribution to speech intelligibility of different envelope modulation rates within the speech dynamic...

The contribution of envelope cues at different rates to intelligibility in a competing-speech task was measured as a function of the short-term envelope level. The target and background mixture was processed using tone vocoders. Envelope signals for each vocoder channel were simultaneously extracted with two low-pass filters, the cutoff frequency of one filter (L) being two octaves below that of the other (H). The envelope from the H filter was used at the peaks and that from the L filter at valleys, or vice versa. This was achieved by cross-fading between the two envelope signals based on a "switching threshold" that was parametrically varied relative to the long-term RMS level of the channel signal. When the cutoff frequencies of the H and L filters were 50 and 12.5 Hz, changes in speech intelligibility occurred mainly when the switching threshold was between -18 and +10 dB. The range was slightly narrower when the cutoff frequencies of the H and L filters were 200 and 50 Hz. Intensity-importance functions for higher-rate envelope modulations suggested that levels ranging from 20 dB below to about 10 dB above the channel RMS level were important, with maximum importance for levels around -5 dB.