%Aigaion2 BibTeX export from Idiap Publications
%Tuesday 03 December 2024 05:34:02 PM

@INPROCEEDINGS{valente:Interspeech2:2008,
         author = {Valente, Fabio and Hermansky, Hynek},
       projects = {Idiap},
          title = {On the Combination of Auditory and Modulation Frequency Channels for ASR applications},
      booktitle = {Interspeech 2008},
           year = {2008},
           note = {IDIAP-RR 08-12},
       crossref = {valente:rr08-12},
       abstract = {This paper investigates the combination of evidence coming from different frequency channels obtained filtering the speech signal at different auditory and modulation frequencies. In our previous work \cite{icassp2008}},
            pdf = {https://publications.idiap.ch/attachments/papers/2008/valente-Interspeech2-2008.pdf},
     postscript = {ftp://ftp.idiap.ch/pub/papers/2008/valente-Interspeech2-2008.ps.gz},
ipdmembership={speech},
}



crossreferenced publications: 
@TECHREPORT{valente:rr08-12,
         author = {Valente, Fabio and Hermansky, Hynek},
       projects = {Idiap},
          title = {On the Combination of Auditory and Modulation Frequency Channels for ASR applications},
           type = {Idiap-RR},
         number = {Idiap-RR-12-2008},
           year = {2008},
    institution = {IDIAP},
           note = {Published in Interspeech 2008},
       abstract = {This paper investigates the combination of evidence coming from different frequency channels obtained filtering the speech signal at different auditory and modulation frequencies. In our previous work \cite{icassp2008}, we showed that combination of classifiers trained on different ranges of {\it modulation} frequencies is more effective if performed in sequential (hierarchical) fashion. In this work we verify that combination of classifiers trained on different ranges of {\it auditory} frequencies is more effective if performed in parallel fashion. Furthermore we propose an architecture based on neural networks for combining evidence coming from different auditory-modulation frequency sub-bands that takes advantages of previous findings. This reduces the final WER by 6.2\\% (from 45.8\\% to 39.6\\%) w.r.t the single classifier approach in a LVCSR task.},
            pdf = {https://publications.idiap.ch/attachments/reports/2008/valente-idiap-rr-08-12.pdf},
     postscript = {ftp://ftp.idiap.ch/pub/reports/2008/valente-idiap-rr-08-12.ps.gz},
ipdmembership={speech},
}