- 學位論文
考慮社群網路之個人化語言模型及行動裝置上的個人化辨識系統
Personalized Language Modeling Considering Social Networks and Personalized Recognizer on Mobile Devices
若您是本文的作者，可授權文章由華藝線上圖書館中協助推廣。
摘要
由於近年來行動裝置的普及，行動裝置又常只有單一位使用者，行動裝置上的個人化語音辨識系統的重要性日益提高。本論文研究之主軸為充分使用行動裝置使用者在社群網路上的個人化語料，考慮其個人化特徵，以及隨時間變動的即時性，發展個人化語言模型及個人化辨識系統。 本論文主要從使用個人化語料來建立個人化語言模型著手，又因為個人化語料之稀疏性，進一步採用社群網路上的語料發展群力型個人化語言模型，並考慮個人化語言模型隨時間變動的特性，發展出可反映即時性的群力型個人化語言模型，也採用了社群網路上使用者的社交互動特徵，來強化個人化語言模型。本論文亦探討了行動裝置上的辨識系統架構，包括使行動裝置在不需與伺服器連線的情況下亦可以進行大字彙連續語音辨識。
並列摘要
Since mobile devices are popular recently and has personal information, the importance of personalized speech recognition system on mobile devices has raised. This thesis aims to make use of personal data of mobile device users on social network. Considering the personal feature and time variation, develop personalized language model and personalized speech recognition system. This thesis starts from building personalized language model using personal data. Since personal data is sparse, continue to make use of data on social network to develop crowd-sourced personalized language model. Considering time-variation of personal language model, develop real-time crowd-sourced personalized language model. Considering social interaction on social network, develop a social-network enhanced personalized language model. This thesis also discusses about speech recognition system on modile devices, including doing LVCSR without internet connection with server.
參考文獻
[1] J. Cohen, “Embedded Speech Recognition Applications in Mobile Phones: Status, Trends, and Challenges”, ICASSP, 2008.
[6] R. Vergin, D. O' Shaughnessy, and A. Farhat, “Generalized mel frequency cepstral coefficients for large-vocabulary speaker-independent continuous-speech recognition”, IEEE Transactions on Speech and Audio Processing, 1999.
[7] O. Viikki and K. Laurila, “Cepstral domain segmental feature vector normali-zation for noise robust speech recognition”, Speech Communication, 1998.
[10] “HTK Speech Recognition Toolkit”, http://htk.eng.cam.ac.uk/
[14] Huggins-Daines, D., “Pocketsphinx: A Free, Real-Time Continuous Speech Recognition System for Hand-Held Devices”, ICASSP 2006.