The design of acoustic models is of vital importance to build a reliable connection between acoustic wave-form and linguistic messages in terms of individual speech units. According to the characteristic of Chinese ph...The design of acoustic models is of vital importance to build a reliable connection between acoustic wave-form and linguistic messages in terms of individual speech units. According to the characteristic of Chinese phonemes, the base acoustic phoneme units set is decided and refined and a decision tree based state tying approach is explored. Since one of the advantages of top-down tying method is flexibility in maintaining a balance between model accuracy and complexity, relevant adjustments are conducted, such as the stopping criterion of decision tree node splitting, during which optimal thresholds are captured. Better results are achieved in improving acoustic modeling accuracy as well as minimizing the scale of the model to a trainable extent.展开更多
In speech recognition, acoustic modeling always requires tremendous transcribed samples, and the transcription becomes intensively time-consuming and costly. In order to aid this labor-intensive process, Active Learni...In speech recognition, acoustic modeling always requires tremendous transcribed samples, and the transcription becomes intensively time-consuming and costly. In order to aid this labor-intensive process, Active Learning （AL） is adopted for speech recognition, where only the most informative training samples are selected for manual annotation. In this paper, we propose a novel active learning method for Chinese acoustic modeling, the methods for initial training set selection based on Kullback-Leibler Divergence （KLD） and sample evaluation based on multi-level confusion networks are proposed and adopted in our active learning system, respectively. Our experiments show that our proposed method can achieve satisfying performances.展开更多
基金Project 60475007 supported by the National Natural Science Foundation of China
文摘The design of acoustic models is of vital importance to build a reliable connection between acoustic wave-form and linguistic messages in terms of individual speech units. According to the characteristic of Chinese phonemes, the base acoustic phoneme units set is decided and refined and a decision tree based state tying approach is explored. Since one of the advantages of top-down tying method is flexibility in maintaining a balance between model accuracy and complexity, relevant adjustments are conducted, such as the stopping criterion of decision tree node splitting, during which optimal thresholds are captured. Better results are achieved in improving acoustic modeling accuracy as well as minimizing the scale of the model to a trainable extent.
基金Acknowledgements This study is supported by the National Natural Science Foundation of China （60705019）, the National High-Tech Research and Development Plan of China （ 2006AA010102 and 2007AA01Z417）, the NOKIA project, and the 111 Project of China under Grant No. 1308004.
文摘In speech recognition, acoustic modeling always requires tremendous transcribed samples, and the transcription becomes intensively time-consuming and costly. In order to aid this labor-intensive process, Active Learning （AL） is adopted for speech recognition, where only the most informative training samples are selected for manual annotation. In this paper, we propose a novel active learning method for Chinese acoustic modeling, the methods for initial training set selection based on Kullback-Leibler Divergence （KLD） and sample evaluation based on multi-level confusion networks are proposed and adopted in our active learning system, respectively. Our experiments show that our proposed method can achieve satisfying performances.
