Subjective comparison and evaluation of speech enhancement algorithms

By Y. Hu, P. Loizou. Speech Communication, 2007

730 citations recorded by the source corpus.

Making meaningful comparisons between the performance of the various speech enhancement algorithms proposed over the years, has been elusive due to lack of a common speech database, differences in the types of noise used and differences in the testing methodology. To facilitate such comparisons, we report on the development of a noisy speech corpus suitable for evaluation of speech enhancement algorithms. This corpus is subsequently used for the subjective evaluation of 13 speech enhancement methods encompassing four classes of algorithms: spectral subtractive, subspace, statistical-model based and Wiener-type algorithms. The subjective evaluation was performed by Dynastat, Inc. using the ITU-T P.835 methodology designed to evaluate the speech quality along three dimensions: signal distortion, noise distortion and overall quality. This paper reports the results of the subjective tests.

Subfields: Computational Mechanics, Signal Processing

View Paper: https://www.semanticscholar.org/paper/1827aa88038cd2dd848f6811897b73c1432d09b1

How researchers see this paper

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

In Computational Mechanics

In Signal Processing

Citations in the corpus

An algorithm that improves speech intelligibility in noise for normal-hearing listeners. (2009, methodology): “These algorithms, however, have been shown to improve primarily the subjective quality of speech rather than speech intelligibility Hu and Loizou, 2007a, 2007b .”
Robustness of speech quality metrics to background noise and network degradations: Comparing ViSQOL, PESQ and POLQA (2013, background): “3, the NOIZEUS conditions can be seen to cluster into two groups, with a gap in the range 2–2.2 on the y-axis (MOS-LQOn).”
Evaluation of Objective Quality Measures for Speech Enhancement (2008, methodology): “Of these two types of distortion, listeners seem to be influenced the most by the speech distortion when making judgments of overall quality [10], [11].”
Subjective Evaluation of Noise Suppression Algorithms in Crowdsourcing (2021, methodology): “Hu et al. [24] estimated the below linear relationship be-tween ( BAK ), ( SIG ), and ( OVRL ) using the NOIZEUS dataset: OV RL MOS = − 0 .”
Speech Quality Assessment (2011, methodology): “These measures were evaluated using the publicly available noisy speech corpus (NOIZEUS(2)), which was used in a comprehensive subjective quality evaluation [72] of 13 different speech enhancement algorithms encompassing four different classes of algorithms: spectral subtractive, subspace, statistic”
Performance Study of a Convolutional Time-Domain Audio Separation Network for Real-Time Speech Denoising (2020, methodology): “Time-frequency domain representations have been used extensively in traditional noise reduction methods, focusing on estimating and removing the spectral amplitude of noise [1, 2].”
A Crowdsourcing Extension of the ITU-T Recommendation P.835 with Validation (2020, methodology): “99, which is similar to the relation [18] estimated with a different dataset.”
The importance of phase in speech enhancement (2011, methodology): “In our experiments we employ the Noizeus speech corpus (Hu and Loizou, 2007).”
ViSQOL: an objective speech quality model (2015, background): “The scatter shows 112 NOIZEUS conditions and 50 P.Sup23 conditions.”
Reasons why current speech-enhancement algorithms do not improve speech intelligibility and suggested solutions (2011, background): “Hu and Loizou [2], [4] showed that algorithms that improve speech quality do not improve speech intelligibility.”