Prosody dependent speech recognition on radio news corpus of American English

International Journal
2006-01-01 14:00
Authors : Ken Chen, Mark Hasegawa-johnson, Aaron Cohen, Sarah Borys, Sung-Suk Kim, Jennifer Cole, Jeung-Yoon Choi

Year : 2006

Publisher / Conference : IEEE Transactions on Audio, Speech, and Language Processing

Volume : 14, issue 1

Page : 232-245

Does prosody help word recognition? This paper proposes a novel probabilistic framework in which word and phoneme are dependent on prosody in a way that reduces word error rates (WER) relative to a prosody-independent recognizer with comparable parameter count. In the proposed prosody-dependent speech recognizer, word and phoneme models are conditioned on two important prosodic variables: the intonational phrase boundary and the pitch accent. An information-theoretic analysis is provided to show that prosody dependent acoustic and language modeling can increase the mutual information between the true word hypothesis and the acoustic observation by exciting the interaction between prosody dependent acoustic model and prosody dependent language model. Empirically, results indicate that the influence of these prosodic variables on allophonic models are mainly restricted to a small subset of distributions: the duration PDFs (modeled using an explicit duration hidden Markov model or EDHMM) and the acoustic-prosodic observation PDFs (normalized pitch frequency). Influence of prosody on cepstral features is limited to a subset of phonemes: for example, vowels may be influenced by both accent and phrase position, but phrase-initial and phrase-final consonants are independent of accent. Leveraging these results, effective prosody dependent allophonic models are built with minimal increase in parameter count. These prosody dependent speech recognizers are able to reduce word error rates by up to 11% relative to prosody independent recognizers with comparable parameter count, in experiments based on the prosodically-transcribed Boston Radio News corpus.
전체 355
65 Domestic Conference 최택성, 문선국, 박영철 "Relative Specific Loudness Histogram 특징벡터를 이용한 음악 장르 분류" in 2007년도 하계종합학술발표회, 2007
64 Domestic Conference 김정근, 박영철, 윤대희 "정현파 모델을 사용한 낮은 전송률에서의 HE-AAC 복원방법" in 2007년도 하계종합학술발표회, 2007
63 Domestic Conference 이동금, 박영철, 윤대희 "저 전송율 MPEG AAC 에서 복호화되는 오디오 음질의 개선에 대한 연구" in 2007년도 하계종합학술발표회, 2007
62 International Conference Yoomi Hur, Young-Choel Park, Dae Hee Youn "A New Structure for Stereo Acoustic Echo Cancellation Based on Binaural Cue Coding" in 122th Convention of Audio Engineering Society, pp.7094, 2007
61 International Conference Jaseong Lee, Young-Cheol Park, Dae Hee Youn, Hong-Goo Kang "A Hybrid Warped Linear Prediction (WLP) AAC Audio Coding Algorithm" in 122th Convention of Audio Engineering Society, pp.7002, 2007
60 International Conference Keun-Sup Lee, Young-Cheol Park, Dae Hee Youn "Fixed-Point Processing Optimization of Mpeg Audio Encoder Using Statistical Model" in 122th Convention of Audio Engineering Society, pp.7023, 2007
59 Domestic Conference 이동금, 박영철, 윤대희 "저 전송율 환경에서의 음질 개선에 대한 연구" in 2007년도 한국음향학회 춘계학술발표대회, 2007
58 International Conference Min-Seok Choi, Chang-Hyun Baik, Young-Cheol Park, Hong-Goo Kang "A Soft-Decision Adaptation Mode Controller for an Efficient Frequency-Domain Generalized Sidelobe Canceller" in ICASSP, 2007
57 International Journal Chang-Heon Lee, Sung-Kyo Jung, Hong-Goo Kang "Applying A Speaker-dependent Speech Compression Technique to Concatenative TTS Synthesizers" in IEEE Transactions on Audio, Speech, and Language Processing, vol.15, 2, pp.632-640, 2007
56 International Journal Seungil Kim, Chungyong Lee, Hong-Goo Kang "Optimum beamformer in correlated source environments" in Journal of Acoustical Society of America, vol.120, issue 6, 2006