5th International Conference on Spoken Language Processing
The aim of our work is to increase the intelligibility of speech in noise by modifying regions of the signal that contain acoustic cues to consonant identity in order to make it more resistant to subsequent degradation. 36 vowel-consonant-vowel stimuli were recorded by four untrained speakers. The vowel onset/offset and consonant constriction/occlusion regions were selectively amplified and stimuli were presented to listeners in a background of noise (0 dB SNR). Enhanced tokens from all speakers were significantly more intelligible than natural tokens and the improvement was greater for the initially least intelligible speakers. Speech material for two speakers was then presented to Japanese and Spanish learners of English and controls. For all groups, the enhanced consonants were more intelligible. Error patterns were related to the 'distance' between the consonantal systems of the listeners' L1 and L2. These results demonstrate the robustness of our enhancement techniques across speaker and listener types.
Bibliographic reference. Hazan, Valerie / Simpson, Andrew / Huckvale, Mark (1998): "Enhancement techniques to improve the intelligibility of consonants in noise : speaker and listener effects", In ICSLP-1998, paper 0487.