Bayesian learning of speech duration models

Abstract

This paper presents the Bayesian speech duration modeling and learning for hidden Markov model (HMM) based speech recognition. We focus on the sequential learning of HMM state duration using quasi-Bayes (QB) estimate. The adapted duration models are robust to nonstationary speaking rates and noise conditions. In this study, the Gaussian, Poisson, and gamma distributions are investigated to characterize the duration models. The maximum a posteriori (MAP) estimate of gamma duration model is developed. To exploit the sequential learning, we adopt the Poisson duration model incorporated with gamma prior density, which belongs to the conjugate prior family. When the adaptation data are sequentially observed, the gamma posterior density is produced with twofold advantages. One is to determine the optimal QB duration parameter, which can be merged in HMMs for speech recognition. The other one is to build the updating mechanism of gamma prior statistics for sequential learning. EM algorithm is applied to fulfill QB parameter estimation. The adaptation of overall HMM parameters can be performed simultaneously. In the experiments, the proposed adaptive duration model improves the speech recognition performance of Mandarin broadcast news and noisy connected digits. The batch and sequential learning are respectively investigated for MAP and QB duration models.

Deriving disyllabic word variants from a Chinese conversational speech corpus

“…In the MCDC8, the duration of ordinary syllables ranges from 15 to 1110 ms (mean, 173 ms) and is equal to 5.78 syllables per second, which is faster than the articulation rate of news reporters in a Chinese Broadcast News corpus (Chien and Huang, 2003). We ran the variant selection algorithm on all disyllabic words in the MCDC8.…”

Comparative Evaluation of Phone Duration Models for Greek Emotional Speech

“…The phone duration modeling approaches are divided in two major categories: The rule-based (Klatt, 1979) and the data-driven methods (Mobius and Santen, 1996;Santen, 1992;Chen et al, 1998;Chien and Huang, 2003;Lazaridis et al, 2007). In the rulebased methods manually produced rules, extracted from experimental studies on large sets of utterances or based on previous knowledge, are utilized for determining the duration of segments.…”

“…Over the last years various statistical methods have been applied in the phone duration modeling task such as, Linear Regression (LR) (Takeda et al, 1989), decisions tree-based models (Mobius and Santen, 1996), Sums-Of-Products (SOP) (Santen, 1992). Artificial Neural Networks (ANN) techniques (Chen et al, 1998), Bayesian models (Chien and Huang, 2003) and instance-based algorithms (Lazaridis et al, 2007) have also been introduced on the phone duration modeling task. Consequently the data-driven approaches offer us the ability to overcome the time consuming labor of the manual extraction of the rules which are needed in the rule-based approaches.…”

Large vocabulary continuous speech recognition for the transcription of Catalan broadcast news and conversations : towards analysis and modelling of acoustic reduction in spontaneous speech

“…Alternatively, besides following the formal transition probability estimation for HMM, the lack of distinct duration modelling for non-stationary SRs may be addressed by SR dependent HMM [Anastasakos et al., ] [Zheng et al, 2003 or the estimation of transition dependent probability distributions modelling discrete duration length [Chien & Huang, 2003].…”