| Presentation | 2016-01-14 Performance evaluation of CRF/HMM-based automatic accent labeling for speech synthesis Rina Mashiko, Tomoki Koriyama, Takao Kobayashi, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | We have proposed an accent type and phrase boundary estimation technique using acoustic and language models represented by HMM and CRF, respectively. It has been shown that naturalness of synthetic speech generated using HMM-based speech synthesis with the accent labels obtained by the proposed technique approaches that with manually annotated accent labels. In this paper, we investigate an optimal choice of weighting factor that determines the balance of acoustic and language models, and examine the effect of using speaker adaptation in the acoustic model training. Then we give a way of obtaining a better acoustic model for accent labeling that leads to more natural sounding synthetic speech. As a result of objective and subjective evaluation, we show that naturalness of synthetic speech can be improved by determining the weighting factor appropriately. Moreover, in the choice of the optimal weighting factor, it is shown that the use of a small amount of speech data of the target speaker would provide a better result. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | accent / speech synthesis / CRF / HMM |
| Paper # | SP2015-85 |
| Date of Issue | 2016-01-07 (SP) |
| Conference Information | |
| Committee | SP |
|---|---|
| Conference Date | 2016/1/14(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Sunpian Kawasaki |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Synthesis, Generation, Prosody, etc. |
| Chair | Kazunori Mano(Shibaura Inst. of Tech.) |
| Vice Chair | Norihide Kitaoka(Tokushima Univ.) |
| Secretary | Norihide Kitaoka(Tokyo City Univ.) |
| Assistant | Takashi Nose(Tohoku Univ.) / Taichi Asami(NTT) |
| Paper Information | |
| Registration To | Technical Committee on Speech |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Performance evaluation of CRF/HMM-based automatic accent labeling for speech synthesis |
| Sub Title (in English) | |
| Keyword(1) | accent |
| Keyword(2) | speech synthesis |
| Keyword(3) | CRF |
| Keyword(4) | HMM |
| 1st Author's Name | Rina Mashiko |
| 1st Author's Affiliation | Tokyo Institute of Technology(Tokyo Tech) |
| 2nd Author's Name | Tomoki Koriyama |
| 2nd Author's Affiliation | Tokyo Institute of Technology(Tokyo Tech) |
| 3rd Author's Name | Takao Kobayashi |
| 3rd Author's Affiliation | Tokyo Institute of Technology(Tokyo Tech) |
| Date | 2016-01-14 |
| Paper # | SP2015-85 |
| Volume (vol) | vol.115 |
| Number (no) | SP-392 |
| Page | pp.pp.1-6(SP), |
| #Pages | 6 |
| Date of Issue | 2016-01-07 (SP) |