Unsupervised Vocabulary Expansion for Automatic Transcription of Broadcast News

Ohtsuki, Katsutoshi; Hiroshima, Nobuaki; Oku, Masaoki; Imamura, Akira

doi:10.1109/icassp.2005.1415290

Cited by 11 publications

(8 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We compared the results yielded by our method with those yielded by three conventional OOV scoring methods: 1) frequency order (FQ), 2) concept base (CB) [14] and 3) latent dirichlet allocation (LDA) [20]. For CB, we constructed a concept base with 850K words of transcription from call centers that are different from centers of evaluation.…”

Section: Oov Selection Accuracymentioning

confidence: 99%

“…Although these factors can be calculated without any manual efforts, it does not focus on relevances between OOV words in relevant documents and target spoken documents. [14] or latent dirichlet allocation (LDA) [20] can evaluate relevances between OOV words in relevant documents and target spoken documents without any manual efforts. They calculate relevances on the basis of topic distributions of each OOV word and target spoken documents.…”

Section: Semantic Similaritymentioning

confidence: 99%

See 1 more Smart Citation

Automatic Vocabulary Adaptation Based on Semantic and Acoustic Similarities

Yamahata

Yamaguchi

Ogawa

et al. 2014

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYRecognition errors caused by out-of-vocabulary (OOV) words lead critical problems when developing spoken language understanding systems based on automatic speech recognition technology. And automatic vocabulary adaptation is an essential technique to solve these problems. In this paper, we propose a novel and effective automatic vocabulary adaptation method. Our method selects OOV words from relevant documents using combined scores of semantic and acoustic similarities. Using this combined score that reflects both semantic and acoustic aspects, only necessary OOV words can be selected without registering redundant words. In addition, our method estimates probabilities of OOV words using semantic similarity and a class-based N-gram language model. These probabilities will be appropriate since they are estimated by considering both frequencies of OOV words in target speech data and the stable class N-gram probabilities. Experimental results show that our method improves OOV selection accuracy and recognition accuracy of newly registered words in comparison with conventional methods.

show abstract

Section: Oov Selection Accuracymentioning

confidence: 99%

Section: Semantic Similaritymentioning

confidence: 99%

Automatic Vocabulary Adaptation Based on Semantic and Acoustic Similarities

Yamahata

Yamaguchi

Ogawa

et al. 2014

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

show abstract

“…In some models, the approach is to expand the lexicon by adding new words or pronunciations. Ohtsuki et al (2005) propose a two-run model where in the first run, the input speech is recognized by the reference vocabulary and relevant words are extracted from the vocabulary database and added thereafter to the reference vocabulary to build an expanded lexicon. Word recognition is done in the second run based on the lexicon.…”

Section: Related Workmentioning

confidence: 99%

Unsupervised Morphology-Based Vocabulary Expansion

Rasooli

Lippincott

Habash

et al. 2014

Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

View full text Add to dashboard Cite

We present a novel way of generating unseen words, which is useful for certain applications such as automatic speech recognition or optical character recognition in low-resource languages. We test our vocabulary generator on seven low-resource languages by measuring the decrease in out-of-vocabulary word rate on a held-out test set. The languages we study have very different morphological properties; we show how our results differ depending on the morphological complexity of the language. In our best result (on Assamese), our approach can predict 29% of the token-based out-of-vocabulary with a small amount of unlabeled training data.

show abstract

“…The experiments are carried out in the framework of ES-TER evaluation campaign and the Google search engine is used to access Web data 1 . All the tests are performed on about 6 hours of French broadcast news from the test corpus of ESTER 2005.…”

Section: Experimental Frameworkmentioning

confidence: 99%

“…Some papers report experiments on a posteriori search of new words in large external databases [1], but such static approaches fail in contemporary document transcription, where topics and named entities are frequently unexpected. Nervertheless, the web constitutes an immense and continuously updated source of language data, in which most of the possible word-sequences are stored.…”

Section: Introductionmentioning

confidence: 99%

On-demand new word learning using world wide web

Oger

Linarès

Béchet

et al. 2008

2008 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

Most of the Web-based methods for lexicon augmenting consist in capturing global semantic features of the targeted domain in order to collect relevant documents from the Web. We suggest that the local context of the out-of-vocabulary (OOV) words contains relevant information. We propose to use the Web to build locally-augmented lexicons which are used in a final local decoding pass. Our experiments confirm the relevance of the Web for the OOV word retrieval. Different methods are proposed to retrieve the hypothesis words. Finally we present the integration of new words in the transcription process based on part-of-speech models. This technique allows to recover 7.6% of the significant OOV words and the accuracy of the system is improved.

show abstract

Unsupervised Vocabulary Expansion for Automatic Transcription of Broadcast News

Cited by 11 publications

References 13 publications

Automatic Vocabulary Adaptation Based on Semantic and Acoustic Similarities

Automatic Vocabulary Adaptation Based on Semantic and Acoustic Similarities

Unsupervised Morphology-Based Vocabulary Expansion

On-demand new word learning using world wide web

Contact Info

Product

Resources

About