Perceptually-motivated Environment-specific Speech Enhancement
ICASSP 2019, May 2019
Abstract
This paper introduces a deep learning approach to enhance speech recordings made in a specific environment. A single neural network learns to ameliorate several types of recording artifacts, including noise, reverberation, and non-linear equalization. The method relies on a new perceptual loss function that combines adversarial loss with spectrogram features. Both subjective and objective evaluations show that the proposed approach improves on state-of-the-art baseline methods.
Links
- Paper preprint
- Listen to audio clips from our experiments.
Citation
Jiaqi Su, Adam Finkelstein, and Zeyu Jin.
"Perceptually-motivated Environment-specific Speech Enhancement."
ICASSP 2019, May 2019.
BibTeX
@inproceedings{Su:2019:PM, author = "Jiaqi Su and Adam Finkelstein and Zeyu Jin", title = "Perceptually-motivated Environment-specific Speech Enhancement", booktitle = "ICASSP 2019", year = "2019", month = may }