Identification of temporal envelope cues in Chinese tone recognition

For tonal languages such as Mandarin Chinese, tone recognition is important for understanding the meaning of words, phrases or sentences. While fundamental frequency carries the most distinctive information for tone recognition, waveform temporal envelope cues can also produce a high level of tone recognition. This study attempts to identify what types of temporal envelope cues contribute to tone recognition and whether these temporal envelope cues are dependent on speakers and vowel contexts. Several signal-correlated-noise stimuli were generated to separate the contribution of three major temporal envelope cues – duration, amplitude contour, and periodicity – to tone recognition. Perceptual results show that the duration cue contributed mostly to discrimination of Tone-3, the amplitude cue contributed mostly to Tone-3 and Tone-4 discrimination, and the periodicity cue contributed to recognition of all tones. However, tone recognition based on temporal envelope cues was highly variable across speakers and vowel contexts. Acoustic analysis of these temporal envelope cues revealed that this variability in tone recognition is directly related to the acoustic variability between the amplitude contour and fundamental frequency contour.