A Non-Intrusive Quality and Intelligibility Measure of Reverberant and Dereverberated Speech

By Tiago H. Falk, Chenxi Zheng, Wai-Yip Geoffrey Chan. IEEE Transactions on Audio, Speech, and Language Processing, 2010

398 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

A summary of the REVERB challenge: state-of-the-art and remaining challenges in reverberant speech processing research (2016, methodology): “The following instrumental measures were employed: frequency-weighted segmental SNR (FWSegSNR) [13], cepstral distance (CD) [13], log-likelihood ratio (LLR) [13], speech-to-reverberation modulation energy ratio (SRMR) [14], and optionally PESQ [15].”
Non-intrusive Speech Quality Assessment Using Neural Networks (2019, methodology): “We also included two non-intrusive measures as a benchmark: the speech-to-reverberation energy ratio (SRMR) [24], and the ITU-T Recommendation P.”
Intrusive and Non-Intrusive Perceptual Speech Quality Assessment Using a Convolutional Neural Network (2019, methodology): “Model RMSE ρ Non-intrusive, CNN + ELM 0.1656 0.9727 Intrusive, CNN + ELM 0.1378 0.9809 0 0.5 1 1.5 Absolute error 0 20 40 60 80 100 [% ] E rr o rs i n g iv en r an g e CNN, intrusive CNN, non-intrusive PESQ, cubic fit POLQA, cubic fit PESQ POLQA SRMR P.563 Figure 6: MOS estimation error distribution”
The reverb challenge: A common evaluation framework for dereverberation and recognition of reverberant speech (2013, methodology): “The SRMR scores are calculated as in [13].”