| Presentation | 2007/10/18 Statistical properties of STRAIGHT spectral variations in POP-song singing Yuri YOSHIDA, Masanori MORISE, Toru TAKAHASHI, Hideki KAWAHARA, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | A new implementation of STRAIGHT spectral estimation based on so-called TANDEM windowing was applied to investigate statistical properties of vowel spectral variations in POP-song singing. The implementation enabled analysis of real world data as a whole. STRAIGHT spectra of singing voice were converted into MFCC filter outputs and MFCC parameters prior to statistical analyses. Principal component analysis of MFCC converted whole data including vowels, consonants and pauses indicated that more than 90.0% of total variation was resided within the first 5 principal components. It was also found that the space spanned by eigenvectors and that by the MFCC basis functions have similar structure. Relatively large overlap of intra-class distance distributions and interclass distance distributions was observed indicating larger spectral variations of singing voice caused by wider range of variability in pitch, loudness, effort and other timbre related attributes. Implications of these results on adopting a vowel based speech conversion method to singing voice conversion are also discussed. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | STRAIGHT / singing voice / spectral envelope / MFCC / principal component analysis |
| Paper # | SP2007-76 |
| Date of Issue |
| Conference Information | |
| Committee | SP |
|---|---|
| Conference Date | 2007/10/18(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) | Statistical properties of STRAIGHT spectral variations in POP-song singing |
| Sub Title (in English) | |
| Keyword(1) | STRAIGHT |
| Keyword(2) | singing voice |
| Keyword(3) | spectral envelope |
| Keyword(4) | MFCC |
| Keyword(5) | principal component analysis |
| 1st Author's Name | Yuri YOSHIDA |
| 1st Author's Affiliation | Faculty of System Engineering, Wakayama University() |
| 2nd Author's Name | Masanori MORISE |
| 2nd Author's Affiliation | Faculty of System Engineering, Wakayama University |
| 3rd Author's Name | Toru TAKAHASHI |
| 3rd Author's Affiliation | Faculty of System Engineering, Wakayama University |
| 4th Author's Name | Hideki KAWAHARA |
| 4th Author's Affiliation | Faculty of System Engineering, Wakayama University |
| Date | 2007/10/18 |
| Paper # | SP2007-76 |
| Volume (vol) | vol.107 |
| Number (no) | 282 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |