Predicting the Degradation of Speech Recognition Performance from Sub-band Dynamic Ranges

• Item type: Journal(1)
• 公開日: 2002-07-15
• タイトル: Predicting the Degradation of Speech Recognition Performance from Sub-band Dynamic Ranges
• タイトル / 言語: en
• タイトル: Predicting the Degradation of Speech Recognition Performance from Sub-band Dynamic Ranges
• 言語: eng
• キーワード / 主題Scheme: Other
• キーワード / 主題: 特集：音声言語情報処理とその応用
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_6501
• 資源タイプ: journal article
• その他タイトル / その他のタイトル: 音声・マルチモーダルインタフェースの実装・評価とその支援
• 著者所属: Graduate School of Engineering Nagoya University
• 著者所属: Center for Integrated Acoustic Information Research Nagoya University
• 著者所属: Center for Integrated Acoustic Information Research Nagoya University
• 著者所属(英): en
• 著者所属(英): Graduate School of Engineering, Nagoya University
• 著者所属(英): en
• 著者所属(英): Graduate School of Engineering, Nagoya University/Center for Integrated Acoustic Information Research, Nagoya University
• 著者所属(英): en
• 著者所属(英): Graduate School of Engineering, Nagoya University/Center for Integrated Acoustic Information Research, Nagoya University
• 著者名: Masato, Kondo; Kazuya, Takeda; Fumitada, Itakura
• 著者名(英): Masato, Kondo; Kazuya, Takeda; Fumitada, Itakura
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: An acoustic measure for predicting the degradation of speechrecognition performance due to noise contamination is developed. Themerits of the proposed measure over using conventional SNR are that 1)the measure does not require original clean signal as a referencesignal 2) the measure takes the spectral shape of noise into accountand 3) the measure can be used to predict recognition performance directly. Thebasic idea of the measure is to utilize the dynamic range of the sub-bandsignals as an estimate of the SNR and to predict the degradation of recognitionperformance by taking the product of the recognition accuracy of eachsub-band. The proposed measure is tested through experimentalevaluation using white Gaussian noise and human-speech-like noise (HSN). Inthe experiment the correlation between the predicted and the actualrecognition accuracies are 0.96 and 0.99 for white noise and HSN respectively. The results confirm the effectiveness of the proposed measure.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: An acoustic measure for predicting the degradation of speechrecognition performance due to noise contamination is developed. Themerits of the proposed measure over using conventional SNR are that 1)the measure does not require original clean signal as a referencesignal, 2) the measure takes the spectral shape of noise into accountand, 3) the measure can be used to predict recognition performance directly. Thebasic idea of the measure is to utilize the dynamic range of the sub-bandsignals as an estimate of the SNR and to predict the degradation of recognitionperformance by taking the product of the recognition accuracy of eachsub-band. The proposed measure is tested through experimentalevaluation using white Gaussian noise and human-speech-like noise (HSN). Inthe experiment, the correlation between the predicted and the actualrecognition accuracies are 0.96 and 0.99 for white noise and HSN,respectively. The results confirm the effectiveness of the proposed measure.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AN00116647
• 書誌情報: 情報処理学会論文誌; 巻 43, 号 7, p. 2242-2248, 発行日 2002-07-15
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 1882-7764