ARTICLE Ganapathy_JASA-EL_2008/IDIAP Modulation Frequency Features For Phoneme Recognition In Noisy Speech Ganapathy, Sriram Thomas, Samuel Hermansky, Hynek EXTERNAL https://publications.idiap.ch/attachments/papers/2008/Ganapathy_JASA-EL_2008.pdf PUBLIC https://publications.idiap.ch/index.php/publications/showcite/Ganapathy_Idiap-RR-70-2008 Related documents Journal of Acoustical Society of America - Express Letters 2008 November 2008 In this letter, a new feature extraction technique based on modulation spectrum derived from syllable-length segments of sub-band temporal envelopes is proposed. These sub-band envelopes are derived from auto-regressive modelling of Hilbert envelopes of the signal in critical bands, processed by both a static (logarithmic) and a dynamic (adaptive loops) compression. These features are then used for machine recognition of phonemes in telephone speech. Without degrading the performance in clean conditions, the proposed features show significant improvements compared to other state-of-the-art speech analysis techniques. In addition to the overall phoneme recognition rates, the performance with broad phonetic classes is reported. REPORT Ganapathy_Idiap-RR-70-2008/IDIAP Modulation Frequency Features For Phoneme Recognition In Noisy Speech Ganapathy, Sriram Thomas, Samuel Hermansky, Hynek EXTERNAL https://publications.idiap.ch/attachments/reports/2008/Ganapathy_Idiap-RR-70-2008.pdf PUBLIC Idiap-RR-70-2008 2008 Idiap October 2008 In this paper, a new feature extraction technique based on modulation spectrum derived from syllable-length segments of sub-band temporal envelopes is proposed. These sub-band envelopes are derived from auto-regressive modelling of Hilbert envelopes of the signal in critical bands, processed by both a static (logarithmic) and a dynamic (adaptive loops) compression. These features are then used for machine recognition of phonemes in telephone speech. Without degrading the performance in clean conditions, the proposed features show significant improvements compared to other state-of-the-art speech analysis techniques. In addition to the overall phoneme recognition rates, the performance with broad phonetic classes is reported.