| Presentation | 2004/12/14 Canonicalization of Feature Parameters targeting Background Noise Takashi FUKUDA, Tsuneo NITTA, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Acoustic models (AMs) of an HMM-based classifier include various types of hidden variables such as gender type, speaking rate, and acoustic environment. If there exists a canonicalization process that reduces the influence of the hidden variables from AMs, a robust automatic speech recognition (ASR) system can be realized, hi this paper, we describe the configuration of a canonicalization process targeting gender type and noise intensity as hidden variables. The proposed canonicalization process is composed of multiple distinctive phonetic feature (DPF) extractors corresponding to the hidden variables and a DPF selector which selects a desirable DPF from multiple DPFs. hi a DPF extraction stage, two approaches, namely (A) extracting DPFs directly from each DPF extractor which represents both gender type and noise intensity, and (B) dividing the DPF extraction stage into a DPF extraction part of gender type and a noise suppression part of different S/N-ratio type, are investigated. Experiments are carried out by comparing the combination of the canonicalized DPF and a single HMM classifier, and also the combination of a single acoustic feature (MFCC) and multiple HMM classifiers. The result shows that the proposed canonicalization method outperformed both of the conventional ASR with MFCC and a single HMM, and the ASR with multiple HMMs |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | automatic speech recognition / feature extraction / canonicalization / distinctive phonetic feature / noise suppressor |
| Paper # | NLC2004-78,SP2004-118 |
| Date of Issue |
| Conference Information | |
| Committee | SP |
|---|---|
| Conference Date | 2004/12/14(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) | Canonicalization of Feature Parameters targeting Background Noise |
| Sub Title (in English) | |
| Keyword(1) | automatic speech recognition |
| Keyword(2) | feature extraction |
| Keyword(3) | canonicalization |
| Keyword(4) | distinctive phonetic feature |
| Keyword(5) | noise suppressor |
| 1st Author's Name | Takashi FUKUDA |
| 1st Author's Affiliation | Graduate School of Engineering, Toyohashi University of Technology() |
| 2nd Author's Name | Tsuneo NITTA |
| 2nd Author's Affiliation | Graduate School of Engineering, Toyohashi University of Technology |
| Date | 2004/12/14 |
| Paper # | NLC2004-78,SP2004-118 |
| Volume (vol) | vol.104 |
| Number (no) | 542 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |