eduzhai > Applied Sciences > Engineering >

A Perceptually-Motivated Approach for Low-Complexity Real-Time Enhancement of Fullband Speech

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 5 pages

Abstract: Over the past few years, speech enhancement methods based on deep learninghave greatly surpassed traditional methods based on spectral subtraction andspectral estimation. Many of these new techniques operate directly in the theshort-time Fourier transform (STFT) domain, resulting in a high computationalcomplexity. In this work, we propose PercepNet, an efficient approach thatrelies on human perception of speech by focusing on the spectral envelope andon the periodicity of the speech. We demonstrate high-quality, real-timeenhancement of fullband (48 kHz) speech with less than 5 of a CPU core.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...