Third International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications (MAVEBA 2003)
The objective and subjective classification of unvoiced stop consonants in varying vowel contexts were studied. The objective classification was based on auditory feature vectors obtained by warped linear prediction (WLP) and vector autoregressive (VAR) models for parameter trajectories. In the case of known vowel the unvoiced consonants were classified 98-100% correctly based on the auditory spectral features of the bursts whereas the VAR models for the parameter (formant) trajectories gave at best only 52- 68% correct classification. The importance of the burst part also in the human perception was confirmed by a listening test.
Index Terms. Speech, syllables, classification
Full Paper (reprinted with permission from Firenze University Press)
Bibliographic reference. Hirvonen, T. / Laine, Unto K. (2003): "Comparison of objective and subjective classification of unvoiced stop consonants in stop-vowel syllables", In MAVEBA-2003, 265-268.