Article details

Research area
Text to speech

Location
Shenzhen, China

Date
2001

Author(s)
Lu Xin, Liu Zhanyi, Zhao Tiejun, Wang Li

Dealing with Polyphone in Text-to-Speech Systems Using How-Net

Synopsis:

Dealing with polyphones is an important part of Chinese
text-to-speech system. Because the pronunciation of a
Chinese character is directly related to the meaning of it,
an algorithm based on semantic calculation using How-Net
is introduced to determine the pronunciations of the
polyphones in new words, which hasn’t appeared in the
polyphone list, the polyphone knowledge base or the
corpus. The experiment results prove it can do good
performance.

Read/download now