Search

This paper details our subword confusion network based approach for Mandarin spoken term detection. As well as the system description, two approaches are presented for improvement of our baseline system. To reduce the inherent high recognition error of the subword decoding system due to its weak language model constraints, the subword confusion network is proposed to be generated from the word decoding system. In addition, a variant of minimum edit distance method (MED) is proposed for linearly scanning the confusion networks for spoken term detection, which incorporates the confidence from confusion networks and other sources. A real-time term detector is constructed based on the modified MED method. Experiments show significant performance improvement from the word decoding and slight improvement from the real-time detector compared to our baseline system.

For a reading tutor, the most important task is to detect the reading miscues such as insertions, omissions, etc. This paper constructed a Chinese reading miscues detection system based on technologies of general large vocabulary continuous speech recognition and proposed two methods to improve the performance of the detection. The first is to align the reference to the confusion network resulted from the recognition instead of the 1-best result to find out the reading miscues. And the second is using the knowledge of reference to regulate the decoding by weighting the n-gram language model probability, if it exists in the reference. The experiments on a Chinese Mandarin reading corpus proved the effectiveness of these two modifications. The detection MDerr and FArate are depressed 50.1% and 70.8% totally by these two methods.