sharma17@interspeech_2017@ISCA

#1 Vowel Onset Point Detection Using Sonority Information [PDF] [Copy] [Kimi¹]

Authors: Bidisha Sharma ; S.R. Mahadeva Prasanna

Vowel onset point (VOP) refers to the starting event of a vowel, that may be reflected in different aspects of the speech signal. The major issue in VOP detection using existing methods is the confusion among the vowels and other categories of sounds preceding them. This work explores the usefulness of sonority information to reduce this confusion and improve VOP detection. Vowels are the most sonorant sounds followed by semivowels, nasals, voiced fricatives, voiced stops. The sonority feature is derived from the vocal-tract system, excitation source and suprasegmental aspects. As this feature has the capability to discriminate among different sonorant sound units, it reduces the confusion among onset of vowels with that of other sonorant sounds. This results in improved detection and resolution of VOP detection for continuous speech. The performance of proposed sonority information based VOP detection is found to be 92.4%, compared to 85.2% by the existing method. Also the resolution of localizing VOP within 10 ms is significantly enhanced and a performance of 73.0% is achieved as opposed to 60.2% by the existing method.