Name
Abstract | ||
|---|---|---|
Standard automatic speech recognition (ASR) systems rely on transcribed speech, language models, and pronunciation dictionaries to achieve state-of-the-art performance. The unavailability of these resources constrains the ASR technology to be available for many languages. In this paper, we propose a novel zero-resourced ASR approach to train acoustic models that only uses list of probable words from the language of interest. The proposed approach is based on Kullback-Leibler divergence based hidden Markov model (KL-HMM), grapheme subword units, knowledge of grapheme-to-phoneme mapping, and graphemic constraints derived from the word list. The approach also exploits existing acoustic and lexical resources available in other resource rich languages. Furthermore, we propose unsupervised adaptation of KL-HMM acoustic model parameters if untranscribed speech data in the target language is available. We demonstrate the potential of the proposed approach through a simulated study on Greek language. |
Year | DOI | Venue |
|---|---|---|
2013 | 10.1109/ASRU.2013.6707771 | ASRU |
Keywords | Field | DocType |
zero-resourced asr,automatic speech recognition systems,unsupervised adaptation,lexical resources,phonemes,speech recognition,greek language,language models,kl-hmm,graphemic constraints,acoustic models,kullback-leibler divergence based hidden markov model,pronunciation dictionaries,resource rich languages,graphemes,grapheme subword units,transcribed speech,grapheme-to-phoneme mapping,probabilistic lexical modeling,untranscribed speech data,natural language processing,hidden markov models,acoustic resources,zero-resourced speech recognition,unsupervised training | Pronunciation,Computer science,Unavailability,Artificial intelligence,Natural language processing,Probabilistic logic,Language model,Pattern recognition,Grapheme,Exploit,Speech recognition,Hidden Markov model,Acoustic model | Conference |
Citations | PageRank | References |
1 | 0.35 | 0 |
Authors | ||
3 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Ramya Rasipuram | 1 | 57 | 6.90 |
| Marzieh Razavi | 2 | 30 | 4.12 |
| Mathew Magimai-Doss | 3 | 516 | 54.76 |