Full metadata record
DC Field | Value | Language |
---|---|---|
dc.contributor.author | 黃培遠 | en_US |
dc.contributor.author | Huang, Pei-Yuan | en_US |
dc.contributor.author | 李其昌 | en_US |
dc.contributor.author | Li, Qi-Chang | en_US |
dc.date.accessioned | 2014-12-12T02:04:03Z | - |
dc.date.available | 2014-12-12T02:04:03Z | - |
dc.date.issued | 1985 | en_US |
dc.identifier.uri | http://140.113.39.130/cdrfb3/record/nctu/#NT744436007 | en_US |
dc.identifier.uri | http://hdl.handle.net/11536/52697 | - |
dc.description.abstract | 本文提出一個通用的連續數字國語語音辨認系統,系統中對於不同的語者收集認音信 號,經過線性預估編碼,二元樹向量量化後,針對每個單字建成隱藏式馬可夫模型。 在辨認的步驟中,將連續語音信號針對每一個建好的單字馬可夫模型,使用分層建立 和凡特比法則來決定待測連續語音是由那幾個單字組成。 本文並針對向量量化的量化碼個數,以及馬可夫模型的狀態數等變數,對於辨認結準 確率的影響加以討論。 | zh_TW |
dc.language.iso | zh_TW | en_US |
dc.subject | 二位樹 | zh_TW |
dc.subject | 向量量化 | zh_TW |
dc.subject | 向量 | zh_TW |
dc.subject | 量化 | zh_TW |
dc.subject | 馬可夫模型 | zh_TW |
dc.subject | 語音 | zh_TW |
dc.subject | 辨認 | zh_TW |
dc.subject | 語音辨認 | zh_TW |
dc.subject | 電信 | zh_TW |
dc.subject | 電子工程 | zh_TW |
dc.subject | TELECOMMUNICATION | en_US |
dc.subject | ELECTRONIC-ENGINEERING | en_US |
dc.title | 利用向量量化和隱藏式馬可夫模型以及分層建立方法的非特定語者國語連續數字語音辨認 | zh_TW |
dc.title | Speaker-independent connected Chinese spoken word recognition based on vector quantization, hidden Markov model and level building | en_US |
dc.type | Thesis | en_US |
dc.contributor.department | 電信工程研究所 | zh_TW |
Appears in Collections: | Thesis |