Effects of noise suppression and envelope dynamic range compression on the intelligibility of vocoded sentences for a tonal language

摘要

Vocoder simulation studies have suggested that the carrier signal type employed affects the intelligibility of vocoded speech. The present work further assessed how carrier signal type interacts with additional signal processing, namely, single-channel noise suppression and envelope dynamic range compression, in determining the intelligibility of vocoder simulations. In Experiment 1, Mandarin sentences that had been corrupted by speech spectrum-shaped noise (SSN) or two-talker babble (2TB) were processed by one of four single-channel noise-suppression algorithms before undergoing tone-vocoded (TV) or noise-vocoded (NV) processing. In Experiment 2, dynamic ranges of multiband envelope waveforms were compressed by scaling of the mean-removed envelope waveforms with a compression factor before undergoing TV or NV processing. TV Mandarin sentences yielded higher intelligibility scores with normal-hearing (NH) listeners than did noise-vocoded sentences. The intelligibility advantage of noise-suppressed vocoded speech depended on the masker type (SSN vs 2TB). NV speech was more negatively influenced by envelope dynamic range compression than was TV speech. These findings suggest that an interactional effect exists between the carrier signal type employed in the vocoding process and envelope distortion caused by signal processing. (C) 2017 Acoustical Society of America.