Presentation | 2008-10-23 An MRHSMM-based voice quality control technique for synthetic speech using speaker adaptation from average voice model Makoto TACHIBANA, Akifumi KOUNO, Takashi NOSE, Takao KOBAYASHI, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | This paper describes a technique for controlling voice quality of synthetic speech using multiple-regression hidden semi-Markov model (MRHSMM). To achieve voice quality control with a small amount of training data, we incorporate a speaker adaptation technique from an average voice model into MRHSMM-based voice quality control. In the proposed technique, we first adapt the average voice model to respective training speakers using a small amount of adaptation data. Then, using obtained speaker-adapted HSMMs and low-dimensional voice quality control vector for each training speaker, the regression matrices of MRHSMM are estimated based on least square method and maximum likelihood estimation. We attempt to control voice quality of synthetic speech using 20 speakers' data of 50 sentences for each speaker. From results of subjective evaluation, we show that the proposed technique can control several voice qualities of synthetic speech. Furthermore, we propose model interpolation technique for the MRHSMMs and show its evaluation results. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | HMM-based speech synthesis / voice quality control / multiple-regression HSMM / average voice model / speaker adaptation |
Paper # | SP2008-63 |
Date of Issue |
Conference Information | |
Committee | SP |
---|---|
Conference Date | 2008/10/16(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | |
Vice Chair | |
Secretary | |
Assistant |
Paper Information | |
Registration To | Speech (SP) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | An MRHSMM-based voice quality control technique for synthetic speech using speaker adaptation from average voice model |
Sub Title (in English) | |
Keyword(1) | HMM-based speech synthesis |
Keyword(2) | voice quality control |
Keyword(3) | multiple-regression HSMM |
Keyword(4) | average voice model |
Keyword(5) | speaker adaptation |
1st Author's Name | Makoto TACHIBANA |
1st Author's Affiliation | Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology() |
2nd Author's Name | Akifumi KOUNO |
2nd Author's Affiliation | Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology |
3rd Author's Name | Takashi NOSE |
3rd Author's Affiliation | Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology |
4th Author's Name | Takao KOBAYASHI |
4th Author's Affiliation | Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology |
Date | 2008-10-23 |
Paper # | SP2008-63 |
Volume (vol) | vol.108 |
Number (no) | 265 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |