Abstract

Speech recognition has only recently been applied to Cantonese. Considerable effort, however, has been spent in recognizing Mandarin, the standard dialect of Chinese. Prior to this thesis, the only published work on monosyllabic Cantonese tone recognition is from Tan Lee et al. (1993,1995). This thesis is the first of its kind in that it explores memory-based learning as a viable approach for Cantonese tone recognition. The memory-based learning algorithm employed in this thesis outperforms the highly respected and widely used neural network approach. Various numbers of tones and features are modeled to find the best method for feature selection and extraction. To further optimize this approach, experiments are performed to isolate the best feature weighting method, best class voting weights method, and the best number of k-values to implement. A detailed error analysis is also reported. This thesis will prove valuable as a future reference for memory-based learning in application to more complex tasks such as continuous speech tone recognition.

Degree

College and Department

Humanities; Linguistics and English Language

Rights

http://lib.byu.edu/about/copyright/