Reverberant speech recognition exploiting clarity index estimation
| Autor: | Patrick A. Naylor, Dushyant Sharma, Toon van Waterschoot, Pablo Peso Parada |
|---|---|
| Přispěvatelé: | Commission of the European Communities |
| Jazyk: | angličtina |
| Rok vydání: | 2015 |
| Předmět: |
Reverberation
Technology Science & Technology Acoustic model selection SISTA business.industry Computer science Speech recognition Feature vector Estimator Word error rate Acoustic model Pattern recognition Engineering Electrical & Electronic Artificial Intelligence And Image Processing Reverberant speech recognition Reduction (complexity) Engineering Electrical And Electronic Engineering Artificial intelligence business Hidden Markov model Baseline (configuration management) Networking & Telecommunications C-50 |
| Popis: | We present single-channel approaches to robust automatic speech recognition (ASR) in reverberant environments based on non-intrusive estimation of the clarity index (C 50). Our best performing method includes the estimated value of C 50 in the ASR feature vector and also uses C 50 to select the most suitable ASR acoustic model according to the reverberation level. We evaluate our method on the REVERB Challenge database employing two different C 50 estimators and show that our method outperforms the best baseline of the challenge achieved without unsupervised acoustic model adaptation, i.e. using multi-condition hidden Markov models (HMMs). Our approach achieves a 22.4 % relative word error rate reduction in comparison to the best baseline of the challenge. ispartof: EURASIP Journal on Advances in Signal Processing vol:2015 issue:1 pages:1-12 status: published |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|