该文提出了一种基于音子HMM输出概率分布(OPD)计算集外词(OOV)拒绝的方法,该方法主要用于语音识别中的验证阶段。与动态垃圾模型中使用经过排序的概率数值的方法相比,OPD向量包含了更多的信息。每个音素的置信值都是以OPD向量为输入的支持向量机(SVM)分别计算出,并得到词的置信度,确定候选词被接受或拒绝。实验结果表明,所提出的方法在语音识别的验证任务中,与传统的动态垃圾模型相比,等错误率EER值相对降低了11.0%。
Abstract
This paper proposes an Out-Of-Vocabulary (OOV) word rejection method based on the Output Probability Distribution (OPD) of phoneme HMMs in word verification. Compared with input vector for dynamic garbage model, OPD vector contains more information than the sorted probabilities. Confidence score of each phoneme is calculated by SVM with OPD vectors as input to determine the acceptance or rejection of the hypotheses. Experimental results show that the proposed method achieved 11.0% decrease in EER than the conventional dynamic garbage model in word verification task.
Key wordsspeech recognition; word verification; confidence measure
关键词
语音识别 /
关键词确认 /
置信度
{{custom_keyword}} /
Key words
speech recognition /
word verification /
confidence measure
/
/
/
/
/
/
/
/
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] C C Broun, W M Campbell, Robust Out-of-Vocabulary Rejection for Low-Complexity Speaker in Dependent Speech Recognition[C]//Proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing 2000, 2000,3: 1811-1814.
[2] Tepperman J, Narayanan S, Using Articulatory Representations to Detect Segmental Errors in Non-native Pronunciation[J], IEEE Trans. On Audio, Speech, and Language Processing, 2011, 19(8): 8-22.
[3] V Vapnik, Statistical Learning Theory[M], New York, U.S., John Wiley and Sons Inc., 1998.
[4] Benayed et al, Confidence measures for keyword spotting using support vector machines[C]//Proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing 2003, 2003, 1: 588-591.
[5] Takehito Utsuro et al, Confidence of agreement among multiple LVCSR models and model combination by SVM[C]//Proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing 2003, 2003, 1, I-16-19.
[6] Haiyang Li, Jiqing Han, Tieran Zheng, et al. Mandarin keyword spotting using syllable based confidence features and SVM[C]//Proceedings of 2nd Intelligent Control and Information Processing (ICICIP), 2011,1: 256-259.
[7] 孙成立,刘刚,郭军. 中文语音确认中子词置信度性能的研究[J].中文信息学报,2008,22(2):106-109.
[8] 孟莎,刘加. 汉语语音检索的集外词问题与两阶段检索方法[J].中文信息学报,2009,23(6):91-97.
[9] Dong Yu, Jinyu Li, Li Deng. Calibration of Confidence Measures in Speech Recognition[J]. IEEE Transactions on Audio, Speech, and Language Processing, 2011, 19(8): 2463-2471.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
基金
国家自然科学基金资助项目(61073017);深圳基础研究重点资助项目(JC201005260245A)
{{custom_fund}}
PDF(1896 KB)
