This paper reports on the participation of FBK at the IWSLT 2011 Evaluation: namely in the English ASR track, the Arabic-English MT track and the English-French MT and SLT tracks. Our ASR system features acoustic models trained on a portion of the TED talk recordings that was au- tomatically selected according to the fidelity of the provided transcriptions. Three decoding steps are performed inter- leaved by acoustic feature normalization and acoustic model adaptation. Concerning the MT and SLT systems, besides language specific pre-processing and the automatic introduc- tion of punctuation in the ASR output, two major improve- ments are reported over our last year baselines. First, we applied a fill-up method for phrase-table adaptation; second, we explored the use of hybrid class-based language models to better capture the language style of public speeches.
FBK @ IWSLT 2011
Bisazza, Arianna;Brugnara, Fabio;Falavigna, Giuseppe Daniele;Giuliani, Diego;Gretter, Roberto;Federico, Marcello
2011-01-01
Abstract
This paper reports on the participation of FBK at the IWSLT 2011 Evaluation: namely in the English ASR track, the Arabic-English MT track and the English-French MT and SLT tracks. Our ASR system features acoustic models trained on a portion of the TED talk recordings that was au- tomatically selected according to the fidelity of the provided transcriptions. Three decoding steps are performed inter- leaved by acoustic feature normalization and acoustic model adaptation. Concerning the MT and SLT systems, besides language specific pre-processing and the automatic introduc- tion of punctuation in the ASR output, two major improve- ments are reported over our last year baselines. First, we applied a fill-up method for phrase-table adaptation; second, we explored the use of hybrid class-based language models to better capture the language style of public speeches.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
