Computer modelling of lexical tone perception

Chen Fangxin, A.J. Rozsypal


Theoretical importance and practical implications of research on computer speech recognition generate considerable interest in this field among linguists, psychologists, and communication engineers alike. The long-term aim is to design a model reflecting the speech perception process and eventually simulate it by a computer program able to perform speaker-independent speech recognition. The present study focuses on one aspect relevant for tone languages, the recognition of lexical tones. The language chosen is Mandarin Chinese (MC). To build the tone-perception model, two aspects of tone were investigated: perceptual dimensions and domain. The perceptual dimensions specify which acoustic cues contribute to the lexical tone perception and how these cues are represented in the auditory system. Tone domain determines where in the syllable the lexical tone is physically located


hearing; speech intelligibility; speech recognition; computer modelling; lexical tone perception; computer speech recognition; speech perception process; speaker-independent speech recognition; Mandarin Chinese; tone-perception model; perceptual dimensions; domain; acoustic cues

Full Text:



  • There are currently no refbacks.