13th Annual Conference of the International Speech Communication Association

Portland, OR, USA
September 9-13, 2012

Combining Multiple High Quality Corpora for Improving HMM-TTS

Vincent Wan, Javier Latorre, K. K. Chin, Langzhou Chen, Mark J. F. Gales, Heiga Zen, Kate Knill, Masami Akamine

Toshiba Research Europe Ltd., Cambridge Research Lab, Cambridge, UK

The most reliable way to build synthetic voices for end-products is to start with high quality recordings from professional voice talents. This paper describes the application of average voice models (AVMs) and a novel application of cluster adaptive training (CAT) to combine a small number of these high quality corpora to make best use of them and improve overall voice quality in hidden Markov model based text-tospeech (HMMTTS) systems. It is shown that integrated training by both CAT and AVM approaches, yields better sounding voices than speaker dependent modelling. It is also shown that CAT has an advantage over AVMs when adapting to a new speaker. Given a limited amount of adaptation data CAT maintains a much higher voice quality even when adapted to tiny amounts of speech.

Index Terms: speech synthesis, cluster adaptive training, speaker adaptation, average voice models

Full Paper

Bibliographic reference.  Wan, Vincent / Latorre, Javier / Chin, K. K. / Chen, Langzhou / Gales, Mark J. F. / Zen, Heiga / Knill, Kate / Akamine, Masami (2012): "Combining multiple high quality corpora for improving HMM-TTS", In INTERSPEECH-2012, 1135-1138.