Variational Mode Decomposition based Emotion Recognition Speech Features from Voiced Regions using Thresholding Technique
Lakshmi Srinivas. D1, Shaik Jakeer Hussain2
1Lakshmi Srinivas D, Department of Electronics and Communication Engineering, VFSTR, Guntur (Andhra Pradesh), India.
2Shaik Jakeer Hussain, Department of Electronics and Communication Engineering, VFSTR, Guntur (Andhra Pradesh), India.
Manuscript received on 07 April 2019 | Revised Manuscript received on 20 April 2019 | Manuscript published on 30 April 2019 | PP: 1460-1467 | Volume-8 Issue-6, April 2019 | Retrieval Number: F3916048619/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Emotion recognition from speech signals is one of the latest research topics involving various emotional speech features for its classification. In this work, a variety of emotional speech features are extracted only from the voiced regions of the emotional speech signal. This algorithm includes the average energy distributed over the frequency range in wavelet domain and zero crossing rate for voiced region detection. The median of ratio of highest sub-band to lowest sub-bands energy and the average zero crossing rate of all segments is considered as thresholds for voiced region detection in speech signal. The voiced regions of speech are filtered to the low frequency range and divided into smaller voiced regions. Intrinsic mode functions and mean frequency components are calculated using Variational mode decomposition (VMD) and Hilbert transform in iterative way. Mean of mean frequencies and mean of inter quartile range of intrinsic mode function of all speech segments are extracted as features, which provide variations in emotional speech for classification. Statistical parameters are calculated on these extracted features only from voiced regions of speech which provide easy process of classification.
Keyword: Speech Classification, Wavelet Transforms, Empirical Mode Decomposition, Intrinsic Mode Functions, Hilbert Huang transform, Variational Mode Decomposition.
Scope of the Article: Pattern Recognition