The objective of the present work is to propose a method to automaticallydetect polarity of the speech signals by estimating instants of significantexcitation of the vocaltract and the cosine phase of the analytic signalrepresentation. The phase changes in the analytic signal around the Hilbertenvelope (HE) peaks are found to vary according to the polarity of the givenspeech signal. The relevant HE peaks for the Hilbert phase analysis areselected by estimating the instants of significant excitation in speech. Thespeech polarity identification rate obtained for the proposed method is almostequal to the state of the art residual skewness method for speech polaritydetection. The proposed method also provides the same results for the polaritydetection in electro-glottogram signals. Finally, the robustness of theproposed method is confirmed from the reduced detection error rates obtained innoisy environments with various signal to noise ratios (SNRs). The MATLAB codesused for implementing the proposed method are available for download from thefollowing link: http://nlp.amrita.edu:8080/TTS/polarityprograms.zip
展开▼
