Adaptation Strategies for the Acoustic and Language Models in Bilingual Speech Transcription

Dieguez-Tirado, Javier; Garcı́a-Mateo, Carmen; Docío-Fernández, Laura; Cardenal-López, Antonio

doi:10.1109/icassp.2005.1415243

Cited by 6 publications

(5 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Para medir a qualidade dos modelos de linguagem construídos com n-gramas extraídos a partir de corpora (Chen & Goodman, 1996;Sennrich, 2012;Dieguez-Tirado et al, 2005) utilizamos a perplexidade:…”

Section: Perplexidadeunclassified

Distância diacrónica automática entre variantes diatópicas do português e do espanhol

Campos¹,

Gamallo²,

Neves

et al. 2020

Linguamática

View full text Add to dashboard Cite

O objetivo deste trabalho é aplicar uma metodologia baseada na perplexidade, para calcular automaticamente a distância interlinguística entre diferentes períodos históricos de variantes diatópicas de idiomas. Esta metodologia aplica-se a um corpus construído adhoc em ortografia original, numa base equilibrada de ficção e não-ficção, que mede a distância histórica entre o português europeu e do Brasil, por um lado, e o espanhol europeu e o da Argentina, por outro. Os resultados mostram distâncias muito próximas em ortografia original e transcrita automaticamente, entre as variedades diatópicas do português e do espanhol, com ligeiras convergências/divergências desde meados do século XX até hoje. É de salientar que o método não é supervisionado e pode ser aplicado a outras variedades diatópicas de línguas.

show abstract

Section: Perplexidadeunclassified

Distância diacrónica automática entre variantes diatópicas do português e do espanhol

Campos¹,

Gamallo²,

Neves

et al. 2020

Linguamática

View full text Add to dashboard Cite

show abstract

“…Perplexity is frequently used as a quality measure for language models built with n-grams extracted from text corpora (Chen and Goodman, 1996;Dieguez-Tirado, Garcia-Mateo, Docio-Fernandez, and Cardenal-Lopez, 2005;Sennrich, 2012). It has also been used in very specific tasks, such as for classifying formal and colloquial tweets (González, 2015), and for identifying closely related languages (Gamallo, Alegria, Pichel, and Agirrezabal, 2016).…”

Section: Perplexity-based Measurementmentioning

confidence: 99%

A Methodology to Measure the Diachronic Language Distance between Three Languages Based on Perplexity

Campos

Gamallo

Alegria³

et al. 2020

Journal of Quantitative Linguistics

View full text Add to dashboard Cite

The aim of this paper is to apply a corpus-based methodology, based on the measure of perplexity, to automatically calculate the cross-lingual language distance between historical periods of three languages. The three historical corpora have been constructed and collected with the closest spelling to the original on a balanced basis of fiction and nonfiction. This methodology has been applied to measure the historical distance of Galician with respect to Portuguese and Spanish, from the Middle Ages to the end of the 20th century, both in original spelling and automatically transcribed spelling. The quantitative results are contrasted with hypotheses extracted from experts in historical linguistics. Results show that Galician and Portuguese are varieties of the same language in the Middle Ages and that Galician converges and diverges with Portuguese and Spanish since the last period of the 19th century. In this process, orthography plays a relevant role. It should be pointed out that the method is unsupervised and can be applied to other languages.

show abstract

“…We employ a large vocabulary continuous speech recognizer based on Continuous Hidden Markov Models (CHMM). The recognition engine is a two-pass recognizer: a Viterbi algorithm, which works in a synchronous way with a beam search, and an A * algorithm [5].…”

Section: Baseline Recognition Systemmentioning

confidence: 99%

“…Our topic-adapted recognizers make use of topic-adapted language models generated from SpeechDAT orthographic transcriptions. Topic adaptation is achieved by mixing n-gram models [5]. This mixture is generated in several steps.…”

Section: Topic-adapted Language Modelsmentioning

confidence: 99%

“…A lower weight (for instance, 5%) will reduce out-of-vocabulary word rate, but the resulting mixed LM will be too general to discriminate between topic utterances and out-of-topic expressions. A more precise solution would select the weights that minimize perplexity for each mixed language model on an evaluation set like done in [5].…”

Section: Topic-adapted Language Modelsmentioning

confidence: 99%

See 1 more Smart Citation

Application of confidence measures for dialogue systems through the use of parallel speech recognizers

López

Mateo²

2005

Interspeech 2005

View full text Add to dashboard Cite

To assess the correctness of a recognizer output in any instance of a dialogue is a complex task that has been studied thoroughly during the past decade. Its importance relays on the need for robust dialogue systems, capable of dealing with difficulties inherent to human-machine communications: user errors and corrections, speech recognizer errors, error recovery techniques, etc.In this paper, we present a novel approach to the problem of deciding what the user has said. We use confidence measures derived from low level knowledge sources (acoustic and linguistic information) and generated in parallel from several topic-adapted speech recognizers. Each recognizer is aimed to the recognition of a particular topic, and confidence measures are compared through the use of a classifier that lead to a most probable solution.This approach shows to be specially suited for difficult topics, such as proper names or confirmations, which are highly meaningful for error correction tasks. These topics present high error rates when using an application-wide speech recognizer, but recognition correction is greatly enhanced through the use of parallel recognizers. Moreover, the use of topic-adapted recognizers seems to help also in the identification of the user intention and in the detection of outof-application utterances.

show abstract

Adaptation Strategies for the Acoustic and Language Models in Bilingual Speech Transcription

Cited by 6 publications

References 9 publications

Distância diacrónica automática entre variantes diatópicas do português e do espanhol

Distância diacrónica automática entre variantes diatópicas do português e do espanhol

A Methodology to Measure the Diachronic Language Distance between Three Languages Based on Perplexity

Application of confidence measures for dialogue systems through the use of parallel speech recognizers

Contact Info

Product

Resources

About