Tree-Based State Tying for High Accuracy Modelling

Tree-Based State Tying for High Accuracy Modelling S.J. Young, J.J. Odell, P.C. Woodland Acoustic Cambridge University Engineering Department Trumpington Street, Cambridge, CB2 1PZ, England ABSTRACT The key problem to be faced when building a HMM-based continuous speech recogniser is maintaining the balance be- tween model complexity and available training data. For large vocabulary systems requiring cross-word context de- pendent modelling, this is particularly acute since many mmh contexts will never occur in the training data. This paper de- scribes a method of creating a tied-state continuous speech recognition system using a phonetic decision tree. This tree- based clustering is shown to lead to similar recognition per- formance to that obtained using an earlier data-driven ap- proach but to have the additional advantage of providing a mapping for unseen triphones. State-tying is also compared with traditional model-based tying and shown to be clearly superior. Experimental results are presented for both the Resource Management and Wall Street 3ournal tasks. 1. INTRODUCTION Hidden Markov Models (HMMs) have proved to be an effective basis for modelling time-varying sequences of speech spectra. However, in order to accurately capture the variations in real speech spectra (both inter-speaker and intra-speaker), it is necessary to have a large number of models and to use relatively complex output probabil- ity distributions. For example, to achieve good perfor- mance in a continuous density HMM system, it is neces- sary to use mixture Gaussian output probability distri- butions together with context dependent phone models. In practice, this creates a data insufficiency problem due to the resulting large number of model parameters. Fur- thermore, the data is usually unevenly spread so that some method is needed to balance model complexity against data availability. This data insufficiency problem becomes acute when a system incorporating cross-word context dependency is used. Because of the large number of possible cross- word triphones, there are many models to estimate and a large number of these triphones will have few, if any, occurrences in the training data. The total number of triphones needed for any particular application depends on the phone set, the dictionary and the grammati- cal constraints. For example, there are about 12,600 position-independent triphones needed for the Resource Management task when using the standard word pair grammar and 20,000 when no grammar is used. For the 20k Wall Street Journal task, around 55,000 triphones are needed. However, only 6600 triphones occur in the Resource Management training data and only 18,500 in the SI84 section of the Wall Street Journal training data. Traditional methods of dealing with these problems in- volve sharing models across differing contexts to form so-called generalised triphones and using a posteriori smoothing techniques[5]. However, model-based sharing is limited in that the left and right contexts cannot be treated independently and hence this inevitably leads to sub-optimal use of the available data. A posteriori smoothing is similarly unsatisfactory in that the models used for smoothing triphones are typically biphones and monophones, and these will be rather too broad when large training sets are …

Content truncated to 500 words