Second Language Studies: Acquisition, Learning, Education and Technology

Tokyo, Japan
September 22-24, 2010

Dialect (Regional Accent)-Based Speaker Classification of Chinese

Xuebin Ma, Nobuaki Minematsu

The University of Tokyo, Japan

In China, different dialects often have different sets of vowels and each dialect comes to have its unique vowel space. In this demo, a participant is asked to read about forty Chinese letters by his/her own dialect and the vowel segments are automatically detected. Using these vowel segments, their vowel space (vowel structure) is estimated based on speech structure theory, where the extra-linguistic factors such as gender and age are effectively removed from speech acoustics and only the linguistic aspect is extracted. Using the vowel structures, Chinese speakers are classified purely based on their dialectal characteristics.

Bibliographic reference.  Ma, Xuebin / Minematsu, Nobuaki (2010): "Dialect (regional accent)-based speaker classification of Chinese", In L2WS-2010, paper D1-7 (abstract).