Contribution of different temporal cues to Cantonese tone perception in quiet and in noise

Abstract

PURPOSE: This study examined the importance of three temporal cues for Cantonese tone identification in quiet and in noise. METHODS: Three temporal cues, namely the amplitude contour (TE50), the periodicity cue (TE500), and the temporal fine structure (TFS), were extracted from 18 familiar words produced in three syllables in the six Cantonese tones. The original stimuli and the extracted cues were combined with two types of noises, including a two-male talker-babble, and speech-shaped noise. Twenty Cantonese-speaking young adults identified the tones in the original stimuli and the stimuli with the three temporal cues in quiet and in the two types of noise. RESULTS: Listeners identified the tones in the original stimuli better than in the three temporal cues in quiet and under noise. Tone recognition in all the three temporal cues was more difficult in two-male talker-babble than in speech-shaped noise. Among the three temporal cues, tone identification accuracy was the highest in TFS, than in TE500, followed by in TE50 in both quiet and noise conditions though the mid-rising tone was identified more accurately in TE500 than in TFS in quiet. For TFS, the accuracy for the identification of tone shapes was similar in quiet and in noise. However, for TE500 and TE50, the accuracy for the identification of tone shapes dropped significantly in noise than in quiet, with the accuracy rates approaching or at chance level. CONCLUSION: Among the three temporal cues, TFS is more important for Cantonese tone perception. Neither TE500 nor TE50 is an effective cue for Cantonese tone perception and both of them are highly susceptible to noise. Masking effect in two-talker babbles is greater than that in speech-shaped noise for Cantonese tone perception, suggesting that voice characteristics in the masker have a detrimental effect on tone perception in noise.