Robust AM-FM features for speech recognition

Dimitriadis, D; Maragos, P; Potamianos, A

dc.contributor.author	Dimitriadis, D	en
dc.contributor.author	Maragos, P	en
dc.contributor.author	Potamianos, A	en
dc.date.accessioned	2014-03-01T01:23:01Z
dc.date.available	2014-03-01T01:23:01Z
dc.date.issued	2005	en
dc.identifier.issn	1070-9908	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/16776
dc.subject	AM-FM	en
dc.subject	ASR	en
dc.subject	Features	en
dc.subject	Nonlinear	en
dc.subject	Speech	en
dc.subject.classification	Engineering, Electrical & Electronic	en
dc.subject.other	Algorithms	en
dc.subject.other	Amplitude modulation	en
dc.subject.other	Feature extraction	en
dc.subject.other	Frequency modulation	en
dc.subject.other	Mathematical models	en
dc.subject.other	Signal filtering and prediction	en
dc.subject.other	Spurious signal noise	en
dc.subject.other	Automatic speech recognition (ASR)	en
dc.subject.other	Mel cepstrum coefficients (MFCCs)	en
dc.subject.other	Speech recognition	en
dc.title	Robust AM-FM features for speech recognition	en
heal.type	journalArticle	en
heal.identifier.primary	10.1109/LSP.2005.853050	en
heal.identifier.secondary	http://dx.doi.org/10.1109/LSP.2005.853050	en
heal.language	English	en
heal.publicationDate	2005	en
heal.abstract	In this letter, a nonlinear AM-FM speech model is used to extract robust features for speech recognition. The proposed features measure the amount of amplitude and frequency modulation that exists in speech resonances and attempt to model aspects of the speech acoustic information that the commonly used linear source-filter model fails to capture. The robustness and discriminability of the AM-FM features is investigated in combination with mel cepstrum coefficients (MFCCs). It is shown that these hybrid features perform well in the presence of noise, both in terms of phoneme-discrimination (J-measure) and in terms of speech recognition performance in several different tasks. Average relative error rate reduction up to 11% for clean and 46% for mismatched noisy conditions is achieved when AM-FM features are combined with MFCCs. © 2005 IEEE.	en
heal.publisher	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC	en
heal.journalName	IEEE Signal Processing Letters	en
dc.identifier.doi	10.1109/LSP.2005.853050	en
dc.identifier.isi	ISI:000231234600007	en
dc.identifier.volume	12	en
dc.identifier.issue	9	en
dc.identifier.spage	621	en
dc.identifier.epage	624	en