Sonority Measurement Using System, Source, and Suprasegmental Information

Abstract

Sonorant sounds are characterized by regions with prominent formant structure, high energy and high degree of periodicity. In this work, the vocal-tract system, excitation source and suprasegmental features derived from the speech signal are analyzed to measure the sonority information present in each of them. Vocal-tract system information is extracted from the Hilbert envelope of numerator of group delay function. It is derived from zero time windowed speech signal that provides better resolution of the formants. A five-dimensional feature set is computed from the estimated formants to measure the prominence of the spectral peaks. A feature representing strength of excitation is derived from the Hilbert envelope of linear prediction residual, which represents the source information. Correlation of speech over ten consecutive pitch periods is used as the suprasegmental feature representing periodicity information. The combination of evidences from the three different aspects of speech provides better discrimination among different sonorant classes, compared to the baseline MFCC features. The usefulness of the proposed sonority feature is demonstrated in the tasks of phoneme recognition and sonorant classification.

Vowel Onset Point Detection Using Sonority Information

“…Apart from sonorants the sonority feature seems to improve performance of VOP detection in case of fricatives, affricates and stops. Moreover, the usefulness sonority feature derived in [17] is explored in case of continuous utterance.The significance of this improvement may be tested in different tasks like consonant-vowel recognition, end point detection and so on.…”

“…These deviations are reflected in the properties of excitation source along with vocal-tract spectrum (VTS) of the produced speech signal [3]. Based on this fact, a 7-dimensional sonority feature is proposed in [17], which has efficacy to discriminate sounds with varying degree of sonority. In the sonority feature, first 5-dimensions represent VTS characteristics, 6 th -dimension represents strength of excitation (SoE) information and 7 th -dimension is for representing suprasegmental behavior.…”

“…These three aspects characterize formant prominence, SoE and periodicity respectively. The feature provides higher values for more sonorant sounds and Copyright © 2017 ISCA lower values for sounds with less spectral and source prominence [17]. Due to the coupling between vocal-tract and excitation source during speech production, it is better to consider the VTS only at glottal closed phase.…”

“…As the existing features for VOP detection do not have the capability to discriminate between different sonorants, in this work an effort is made to detect VOPs using sonority information. The sonority feature proposed in [17] is based on the study of difference in production behavior of different sonorant sounds and its reflection on vocal-tract spectrum, excitation source, suprasegmental behavior of speech signal. Therefore, it has the ability to bring out difference among sonorants.…”

Multi-Level Adaptive Speech Activity Detector for Speech in Naturalistic Environments

“…The vowel-like regions belong to high SNR portion of speech signals and are less affected by noise [17][18][19]. Similarly, glottal activity detection and sonorant region detection are performed to identify the speech regions in a noisy scenario [20,21]. The study in [22] suggests a self-adaptive method for SAD that has been useful.…”