Acoustics-only based automatic phonetic baseform generation

Ramabhadran, Bhuvana; Bahl, Lalit R.; deSouza, P.V.; Padmanabhan, M.

doi:10.1109/icassp.1998.674429

Cited by 24 publications

(9 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…guerilla g ax r ax l ax g ax r ih l ax guerilla g w eh r ih l ax tornados t er n ey d ow z t er n ey d ow z tornados t ao r n ey d ow s t ow r n ey d ow z phone network to select the pronunciation. Additional resources are typically required, including existing pronunciation lexica [14], speech samples [19], [20], linguistic rules [21], or a combination of these. The focus of previous work has been on pronunciation variation [14], [19] or on common words [15], [17].…”

Section: B Methods For Selecting a Pronunciation Modelmentioning

confidence: 99%

Likelihood-Based Semi-Supervised Model Selection With Applications to Speech Processing

White

Khudanpur

Wolfe

2010

IEEE J. Sel. Top. Signal Process.

View full text Add to dashboard Cite

In conventional supervised pattern recognition tasks, model selection is typically accomplished by minimizing the classification error rate on a set of so-called development data, subject to ground-truth labeling by human experts or some other means. In the context of speech processing systems and other large-scale practical applications, however, such labeled development data are typically costly and difficult to obtain. This article proposes an alternative semi-supervised framework for likelihood-based model selection that leverages unlabeled data by using trained classifiers representing each model to automatically generate putative labels. The errors that result from this automatic labeling are shown to be amenable to results from robust statistics, which in turn provide for minimax-optimal censored likelihood ratio tests that recover the nonparametric sign test as a limiting case. This approach is then validated experimentally using a state-of-the-art automatic speech recognition system to select between candidate word pronunciations using unlabeled speech data that only potentially contain instances of the words under test. Results provide supporting evidence for the utility of this approach, and suggest that it may also find use in other applications of machine learning.

show abstract

Section: B Methods For Selecting a Pronunciation Modelmentioning

confidence: 99%

Likelihood-Based Semi-Supervised Model Selection With Applications to Speech Processing

White

Khudanpur

Wolfe

2010

IEEE J. Sel. Top. Signal Process.

View full text Add to dashboard Cite

show abstract

“…Many G2P systems are presented in the literature. Several names are attributed to this task: grapheme-to-phoneme conversion [24,17], phonetic pronunciation modeling [25], letter-to-sound translation [26], letter-to-phoneme conversion [27,7], phonetic baseform generation [28,29], phonetic transcription [30], text-tophoneme mapping [31], among others.…”

Section: Related Workmentioning

confidence: 99%

Improving recognition of proper nouns in ASR through generating and filtering phonetic transcriptions

Laurent

Meignier

Deléglise

2014

Computer Speech & Language

View full text Add to dashboard Cite

Accurate phonetic transcription of proper nouns can be an important resource for commercial applications that embed speech technologies, such as audio indexing and vocal phone directory lookup. However, accurate phonetic transcription is more difficult to obtain for proper nouns than for regular words. Indeed, phonetic transcription of a proper noun depends on both the origin of the speaker pronouncing it and the origin of the proper noun itself.This work proposes a method that allows the extraction of phonetic transcriptions of proper nouns using actual utterances of those proper nouns, thus yielding transcriptions based on practical use instead of mere pronunciation rules.The proposed method consists in a process that first extracts phonetic transcriptions, and then iteratively filters them. In order to initialize the process, an alignment dictionary is used to detect word boundaries. A rule-based graphemeto-phoneme (G2P) generator (LIA PHON [1]), a knowledge-based approach (JSM [2]), and a Statistical Machine Translation (SMT)-based system [9] were evaluated for this alignment. As a result, on the ESTER 1 French broadcast news corpus, we were able to obtain a decrease of the Word Error Rate (WER) on segments of speech with proper nouns, without negatively affecting the WER on the rest of the corpus.

show abstract

“…In standard procedures like those presented in [7] and [8] for example, the speech utterance of the new word is aligned with speaker-independent allophone models. Effectively a decoding from utterance to allophone sequence is performed, with a bigram model on allophones functioning as the language model on words does in a normal utterance-to-text decoding; we will refer to this as the transition model.…”

Section: A Dynamic Vocabularymentioning

confidence: 99%

“…As in a typical speech-to-text decoding, a weighted combination of acoustic model and transition model logprobabilities is used to determine the best allophone sequence. However, it differs from the earlier approaches of [7] and [8] in that the speaker-independent allophone models and the transition model are assigned a weight The advantage of this approach is twofold. First, since we have to deduce the pronunciation of the enrolled words from just one or two speech examples, we may as well use multiple guesses to maximize the chance that one of them will be right.…”

Section: B Automatic Generation Of Multiple Pronunciationsmentioning

confidence: 99%

A robust high accuracy speech recognition system for mobile applications

Deligne

Dharanipragada

Gopinath

et al. 2002

IEEE Trans. Speech Audio Process.

View full text Add to dashboard Cite

This paper describes a robust, accurate, efficient, low-resource, medium-vocabulary, grammar-based speech recognition system using Hidden Markov Models for mobile applications. Among the issues and techniques we explore are improving robustness and efficiency of the front-end, using multiple microphones for removing extraneous signals from speech via a new multi-channel CDCN technique, reducing computation via silence detection, applying the Bayesian information criterion (bic) to build smaller and better acoustic models, minimizing finite state grammars, using hybrid maximum likelihood and discriminative models, and automatically generating baseforms from single new-word utterances.

show abstract

Acoustics-only based automatic phonetic baseform generation

Cited by 24 publications

References 7 publications

Likelihood-Based Semi-Supervised Model Selection With Applications to Speech Processing

Likelihood-Based Semi-Supervised Model Selection With Applications to Speech Processing

Improving recognition of proper nouns in ASR through generating and filtering phonetic transcriptions

A robust high accuracy speech recognition system for mobile applications

Contact Info

Product

Resources

About