Detection of Glottal Excitation Epochs in Speech Signal Using Hilbert Envelope

Hirak Dasgupta, Prem C. Pandey, K S Nataraj

A technique, suitable for real-time processing, is presented for detection of glottal excitation epochs in voiced speech. It uses Hilbert envelope to enhance saliency of the glottal excitation epochs and to reduce the ripples due to the vocal tract filter. The processing comprises the steps of dynamic range compression, calculation of the Hilbert envelope and epoch marking. The first step reduces amplitude variation by applying A-law on the signal envelope. The second step calculates the Hilbert envelope using the output of an FIR filter-based Hilbert transformer and the delay-compensated signal. The third step uses a dynamic peak detector with fast rise and slow fall and nonlinear smoothing using a two-step median-mean filter to further enhance the saliency of the epochs, followed by a differentiator to mark them. The technique is tested using the CMU-ARCTIC database with simultaneously recorded speech and EGG signals. The results showed a good match in the performance of the proposed technique with those of the state-of-the-art techniques and its robustness against highpass filtering. It may be useful for diagnosis of voice disorders and high-quality voice conversion.

 DOI: 10.21437/Interspeech.2018-2014

Cite as: Dasgupta, H., Pandey, P.C., Nataraj, K.S. (2018) Detection of Glottal Excitation Epochs in Speech Signal Using Hilbert Envelope. Proc. Interspeech 2018, 2132-2136, DOI: 10.21437/Interspeech.2018-2014.

  author={Hirak Dasgupta and Prem C. Pandey and K S Nataraj},
  title={Detection of Glottal Excitation Epochs in Speech Signal Using Hilbert Envelope},
  booktitle={Proc. Interspeech 2018},