Unified framework for acoustic topology modelling: ML-SSS and question-based decision trees
نویسندگان
چکیده
State-shared, context-dependent, acoustic HMM's are the basis of practically all large-vocabulary state-of-the-art speech recognition systems. The topology, i.e. state-sharing, is usually trained by decision tree based clustering of similar phonetic contexts, i.e. divisive clustering on the state level. In this paper, we show that Phonetic Decision Trees (PDT) and Maximum Likelihood Successive State Splitting (ML-SSS) can be regarded as variants of the same fundamental partitioning algorithm: the main di erence being that in ML-SSS all possible phoneme combination sets are allowed, whereas in PDT the possible phoneme combination sets are limited based on phonological information that has been decided a-priori and heuristically. A combination of PDT and ML-SSS outperformed both PDT and ML-SSS on a non-read Japanese speech recognition task. To solve the problem of unseen contexts occurring in ML-SSS, the Split History Backo algorithm is introduced.
منابع مشابه
$L$-enriched topological systems---a common framework of $L$-topology and $L$-frames
Employing the notions of the strong $L$-topology introduced by Zhangand the $L$-frame introduced by Yao and the concept of $L$-enrichedtopological system defined in the present paper, we constructadjunctions among the categories {bf St$L$-Top} of strong$L$-topological spaces, {bf S$L$-Loc} of strict $L$-locales and{bf $L$-EnTopSys} of $L$-enriched topological systems. All of theseconcepts are ...
متن کاملLocal Codebook Features for Mono- and Multilingual Acoustic Phonetic Modelling
In this article we present an alternative method for defining the question set used for the induction of acoustic phonetic decision trees. The method is data driven and employs local similarities between the probability density functions of hidden Markov models. We apply the method to monoand multilingual acoustic phonetic modelling, showing that comparable results to the standard method, using...
متن کاملAcoustic Phonetic Modelling using Local Codebook Features
In this article we present an alternative method for defining the question set used for the induction of acoustic phonetic decision trees. The method is data driven and employs local similarities between the probability density functions of hidden Markov models. The method is shown to work at least as well as the standard method using question sets devised by human experts.
متن کاملAn Integrated Enterprise Resources Planning (ERP) Framework forFlexible Manufacturing SystemsUsing Business Intelligence (BI)Tools
Nowadays Business intelligence (BI) tools provide optimal decision making, analyzing, controlling and monitoring of operations in enterprise systems like enterprise resource planning (ERP) and mainly refer to strong decision making methods used in online analytical processing, reporting and data analysis, such as improve internal processes, analysis of resources, information needs analysis, red...
متن کاملContinuous local codebook features for multi- and cross-lingual acoustic phonetic modelling
In this paper we present a method for defining the question set for the induction of acoustic phonetic decision trees. The method is data driven resulting in a continuous feature space in contrast to the usual categorical one. We apply the features to a multilingual speech recognition task, outperforming consistently the standard method using IPA-based characteristics. An extension to cross-lin...
متن کامل