| Presentation | 2012-12-20 Automatic Vocabulary Adaptation for Speech Recognition based on Semantic Similarity and Confidence Measure Shoko Yamahata, Yoshikazu Yamaguchi, Atsunori Ogawa, Hirokazu Masataki, Osamu Yoshioka, Satoshi Takahashi, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Out-Of-Vocabulary utterances are an unavoidable problem in speech recognition systems. And therefore, automatic vocabulary adaptation methods, which detect OOV words from relevant documents and register them with proper probability is an important technique. To improve recognition accuracy of OOV words, our method selects only relevant OOV words with target spoken documents, and prevents recognition error caused by irrelevant OOV words. We use semantic and acoustic similarity between each OOV word and spoken documents to select relevant OOV words. Furthermore, we propose proper probability estimation method for each OOV word using class language models and semantic similarity. Experimental shows that our method improves OOV word selection accuracy, and OOV word recognition accuracy about 5% in F-measure. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Speech Recognition / Out-Of-Vocabulary / Vocabulary Adaptation / Semantic Similarity / Confidence Measure |
| Paper # | SP2012-85 |
| Date of Issue |
| Conference Information | |
| Committee | SP |
|---|---|
| Conference Date | 2012/12/13(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) | Automatic Vocabulary Adaptation for Speech Recognition based on Semantic Similarity and Confidence Measure |
| Sub Title (in English) | |
| Keyword(1) | Speech Recognition |
| Keyword(2) | Out-Of-Vocabulary |
| Keyword(3) | Vocabulary Adaptation |
| Keyword(4) | Semantic Similarity |
| Keyword(5) | Confidence Measure |
| 1st Author's Name | Shoko Yamahata |
| 1st Author's Affiliation | NTT Media Intelligence Laboratories() |
| 2nd Author's Name | Yoshikazu Yamaguchi |
| 2nd Author's Affiliation | NTT Media Intelligence Laboratories |
| 3rd Author's Name | Atsunori Ogawa |
| 3rd Author's Affiliation | NTT Media Communication Science Laboratiories |
| 4th Author's Name | Hirokazu Masataki |
| 4th Author's Affiliation | NTT Media Intelligence Laboratories |
| 5th Author's Name | Osamu Yoshioka |
| 5th Author's Affiliation | NTT Media Intelligence Laboratories |
| 6th Author's Name | Satoshi Takahashi |
| 6th Author's Affiliation | NTT Media Intelligence Laboratories |
| Date | 2012-12-20 |
| Paper # | SP2012-85 |
| Volume (vol) | vol.112 |
| Number (no) | 369 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |