Very Noisy IEEE Utterances

Generated by Ashutosh Pandey during his dissertation study. If you make use of this dataset, please quote the following paper:

• Pandey A. and Wang D.L. (2022): Self-attending RNN for speech enhancement to improve cross-corpus generalization. IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 30, pp. 1374-1385.
  
  

Description

This dataset consists of pairs of clean and noisy utterances for two IEEE speakers, one male and one female, mixed with babble and cafeteria noises at signal-to-noise ratios (SNRs) of -5 dB and -2 dB. They are available in waveform, and can be downloaded in one ZIP file.