Hodnocení kvality řečových signálů na bázi virtuální instrumentace

Abstract

In this bachelor thesis, a program for objective and subjective evaluation of speech quality was created. This program was created in the programming and development environment LabVIEW. Using this environment, a lot of speech recordings and various indoor and outdoor noises were processed. As the objective methods of speech quality assessment, the basic method of global signal-to-noise ratio (GSNR) and the segmental signal-to-noise ratio (SSNR) with an intensive voice activity detector (VAD) were chosen. As a subjective speech evaluation method, the Dynamic time warping (DTW) method was selected, with PARCOR coefficients being chosen as symptom vectors. To filter out smudged recordings, adaptive filtering, LMS and RLS algorithms were used. The previously mentioned objective and subjective methods of speech quality assessment are described in the thesis.