A Study on Speaker-Adaptive Speech Recognition

A Study on Speaker-Adaptive Speech Recognition X.D. Huang School of Computer Science Carnegie Mellon University Pittsburgh, PA 15213 ABSTRACT Speaker-independent system is desirable in many applications where speaker-specific data do not exist. However, if speaker- dependent data are available, the system could be adapted to the specific speaker such that the error rate could be signifi- cantly reduced. In this paper, DARPA Resource Management task is used as the domain to investigate the performance of speaker-adaptive speech recognition. Since adaptation is based on speaker-independent systems with only limited adap- tation data, a good adaptation algorithm should be consistent with the speaker-independent parameter estimation criterion, and adapt those parameters that are less sensitive to the limited training data. Two parameter sets, the codebook mean vector and the output distribution, are regarded to be most important. They are modified in the framework of maximum likelihood estimation criterion according to the characteristics of each speaker. In order to reliably estimate those parameters, output distributions are shared with each other if they exhibit certain acoustic similarity. In addition to modify these parameters, speaker normalization with neural networks is also studied in the hope that acoustic data normalization will not only rapidly adapt the system but also enhance the robustness of speaker- independent speech recognition. Preliminary results indicate that speaker differences can be well minimized. In compar- ison with speaker-independent speech recognition, the error rate has been reduced from 4.3% to 3.1% by only using pa- rameter adaptation techniques, with 40 adaptation sentences for each speaker. When the number of speaker adaptation sentences is comparable to that of speaker-dependent train- ing, speaker-adaptive recognition works better than the best speaker-dependent recognition results on the same test set, which indicates the robustness of speaker-adaptive speech recognition. 1 INTRODUCTION Speaker-independent speech recognition systems could pro- vide users with a ready-to-use system [1, 2, 3, 4]. There is no need to collect speaker-specific data to train the system, but collect data from a variety of speakers to reliably model many different speakers. Speaker-independent systems are defi- nitely desirable in many applications where speaker-specific data do not exist. On the other hand, if speaker-dependent data are available, the system could be adapted to a specific speaker to further reduce the error rate. The problem of speaker-dependent systems is that for large-vocabulary con- tinuous speech recognition, half an hour of speech from the specific speaker is generally needed to reliably estimate sys- tem parameters. The problem of speaker-independent systems is that the error rate of speaker-independent speech recogni- tion systems is generally two to three times higher than that of speaker-dependent speech recognition systems [2, 3]. A logical compromise for a practical system is to start with a speaker-independent system, and then adapt the system to each individual user. Since adaptation is based on the speaker-independent sys- tem with only limited adaptation data, a good adaptation al- gorithm should be consistent with speaker-independent pa- rameter estimation criterion, and adapt those parameters that are less sensitive to the limited training data. Two …

Content truncated to 500 words