Results on VBD testset

Citation Author(s):
Yan
Tongjia
Submitted by:
Yan Tongjia
Last updated:
Mon, 05/13/2024 - 09:25
DOI:
10.21227/tg91-k697
License:
17 Views
Categories:
Keywords:
0
0 ratings - Please login to submit your rating.

Abstract 

This dataset is the result on Voicebank+Demand testset by our proposed model in paper "MDCTCRN: A Lightweight CRN For Monaural Speech Enhancement with Modified DCT" The overall averge PESQ on Voicebank + Demand is 3.14.
This dataset consists of two zip files, one zip file contains the samples with PESQ below 2.6 after noise reduction in the VBD test set, which can reflect the performance of the model in the case of low signal-to-noise ratio, and the other zip file contains all the samples in the VBD test set. All of the above samples are the result of inference from the highest performing model in the paper.
This dataset is a sample result of a test set for a paper submitted to IEEE signal processing letters,which only contains noisy audiofiles and enhanced audiofiles.

Instructions: 

This dataset consists of two zip files, one zip file contains the samples with PESQ below 2.6 after noise reduction in the VBD test set, which can reflect the performance of the model in the case of low signal-to-noise ratio, and the other zip file contains all the samples in the VBD test set. All of the above samples are the result of inference from the highest performing model in the paper.