| Presentation | 2015-10-15 Statistical singing voice conversion based on direct waveform modification and its parameter generation algorithms Kazuhiro Kobayashi, Tomoki Toda, Satoshi Nakamura, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | This report presents a novel statistical singing voice conversion (SVC) technique with direct waveform modification based on the spectrum differential that can convert voice timbre of a source singer into that of a target singer without vocoder-based waveform generation if F0 conversion is not needed between those speakers. SVC makes it possible to convert voice timbre by statistically modeling the relationship between acoustic features of the source and the target singers. However, speech quality of the converted singing voice is significantly degraded compared to that of a natural singing voice due to some factors, such as errors in excitation modeling based on vocoder and over-smoothing effect of a converted spectral parameter trajectory. To address this issue, we propose a statistical conversion process that estimates the spectral differential between the source and target singers' singing voices and directly modifies the source singer's singing voice waveform by filtering it using the estimated spectral differential without using the vocoder-based waveform generation. Furthermore, we also propose a spectral differential parameter estimation method considering global variance to alleviate the over-smoothing effect. We conduct subjective and objective evaluations, demonstrating that the proposed method makes it possible to significantly improve speech quality while preserving the conversion accuracy of singer identity compared to the conventional SVC. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | singing voice conversion / vocoder / direct waveform modification with spectral differential / global variance / Gaussian mixture model |
| Paper # | SP2015-60 |
| Date of Issue | 2015-10-08 (SP) |
| Conference Information | |
| Committee | SP |
|---|---|
| Conference Date | 2015/10/15(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Kobe Univ. |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Speech interface, Synthesis, Dialogue, Application system, 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) | Statistical singing voice conversion based on direct waveform modification and its parameter generation algorithms |
| Sub Title (in English) | |
| Keyword(1) | singing voice conversion |
| Keyword(2) | vocoder |
| Keyword(3) | direct waveform modification with spectral differential |
| Keyword(4) | global variance |
| Keyword(5) | Gaussian mixture model |
| 1st Author's Name | Kazuhiro Kobayashi |
| 1st Author's Affiliation | Nara Institute of Science and Technology(NAIST) |
| 2nd Author's Name | Tomoki Toda |
| 2nd Author's Affiliation | Nara Institute of Science and Technology(NAIST) |
| 3rd Author's Name | Satoshi Nakamura |
| 3rd Author's Affiliation | Nara Institute of Science and Technology(NAIST) |
| Date | 2015-10-15 |
| Paper # | SP2015-60 |
| Volume (vol) | vol.115 |
| Number (no) | SP-253 |
| Page | pp.pp.7-12(SP), |
| #Pages | 6 |
| Date of Issue | 2015-10-08 (SP) |