Relative Importance of Temporal Envelope and Fine Structure Cues in Low- and High-Order Harmonic Regions for Mandarin Lexical-Tone Recognition

摘要

Importance of speech temporal envelope (TE) and fine structure (TFS) cues for lexical-tone recognition has been investigated in normal-hearing subjects[1][2]. The present study explores the relative importance of TE and TFS cues in low- (LH) versus high-order harmonic (HH) regions, using "acoustic chimeras" [3] with Mandarin monosyllables divided into 8 and 16 frequency channels that the current multichannel cochlear prosthesis can provide. The results show: (1) TE in both LH and HH regions make contributions to lexical-tone recognition without the existence of TFS of the original speech, but their relative importance is modulated by the number of channels; (2) TFS in LH region takes the major role in recognition, but is not. enough for perfect performances; (3) TE in both LH and HH regions and TFS in HH region make significant but different complementary contributions based on the presence of TFS in LH region. Current results further address potential implications for cochlear implant stimulations for lexical-tones with combination of newly-developed encoding strategies [4][5][6].