Exhibition Program 13
Science of Communication and Computation
Anytime, anywhere, we can speak like a native!
Speech rhythm conversion by mobile application
Most native Japanese speakers have difficulty speaking English and therefore cannot communicate well in English with native English speakers. We have proposed a method ‘Speak like a native’ that can convert the speaking rhythm of English sentences spoken by native Japanese speakers into stress-timed rhythm by a native English speaker. However, our previous speaking rhythm conversion technique needed the same sentences to be spoken by a native speaker. In this study, we devised rules for converting the speaking rhythm of native Japanese speakers into that of native English speakers using English speech corpus and showed a speaking rhythm conversion of arbitrary speech sentences automatically. Moreover, we developed a conversion system using a mobile device. We hope that this technique will eventually alleviate the burden involved in communication using non-native languages.
-  S. Hiroya, T. Mochida, “Phase equalization-based autoregressive model of speech signals,” in Proc. Interspeech, pp. 42-45, 2010.
 S. Hiroya, “Non-negative temporal decomposition of speech parameters by multiplicative update rules,” IEEE Trans. Audio, Speech, and Lang. Process., Vol. 21, No. 10, pp. 2108-2117, 2013.
 S. Hiroya, “Speaking rhythm conversion of arbitrary speech using a non-negative temporal decomposition and machine learning,” in Proc. Acoustical Society of Japan, Autumn meeting, pp. 423-424, 2017.