完整後設資料紀錄
DC 欄位 | 值 | 語言 |
---|---|---|
dc.contributor.author | 盧維藩 | en_US |
dc.contributor.author | LU, WEI-FAN | en_US |
dc.contributor.author | 魏哲和 | en_US |
dc.contributor.author | 陳信宏 | en_US |
dc.contributor.author | WEI, ZHE-HE | en_US |
dc.contributor.author | CHEN, XIN-HONG | en_US |
dc.date.accessioned | 2014-12-12T02:03:09Z | - |
dc.date.available | 2014-12-12T02:03:09Z | - |
dc.date.issued | 1984 | en_US |
dc.identifier.uri | http://140.113.39.130/cdrfb3/record/nctu/#NT732436013 | en_US |
dc.identifier.uri | http://hdl.handle.net/11536/52111 | - |
dc.description.abstract | 本文提出一國語單字語音分段法:運用語音中的四種特徵(feature)--通過零點 率(zwro crossing rate),能量的對數值(log energy),將一國語單音字分割成 ”無聲音””前母音”,”中後母音”及”鼻音”等四段。 本文劑提出一應用分段法之特定語者國語單字辨認系統。此系統依據分段結果,對預 存之辨認字模(pattern)作先行之分類,以減少須分辨之字數,再應用分段之時軸 線性對應(segment linear timre warping )以決定字模之相似程度。以十個中文 數字(零一九)作測試,對四位發音者,本系統可得平均95%之正確辨認率。 | zh_TW |
dc.language.iso | zh_TW | en_US |
dc.subject | 對數值 | zh_TW |
dc.subject | 極點頻率 | zh_TW |
dc.subject | 特徵 | zh_TW |
dc.subject | 通過零點率 | zh_TW |
dc.subject | 等化殘餘能量 | zh_TW |
dc.subject | 字模 | zh_TW |
dc.subject | 分段時軸線性對應 | zh_TW |
dc.subject | LOG-ENERGY | en_US |
dc.subject | POLE-FREQUCNCE | en_US |
dc.subject | FEATURE | en_US |
dc.subject | ZERO-CROSSING-RATE | en_US |
dc.subject | NORMALIZED-RESIDUAL-ENERGY | en_US |
dc.subject | PATTERN | en_US |
dc.subject | SLTW | en_US |
dc.title | 應用分段法與時軸線性對應之國語單字辨認 | zh_TW |
dc.type | Thesis | en_US |
dc.contributor.department | 電信工程研究所 | zh_TW |
顯示於類別: | 畢業論文 |