Colab adaptation of AudioSR, with some tweaks:
- added a chunking feature to process input of any length
- added stereo handling, stereo input channels will be splitted and processed independantly (dual mono) and then reconstructed as stereo audio.
- added overlap feature to smooth the transitions between chunks (don't use high values because AudioSR is not 100% phase accurate and this will create weird phase cancellations accross the overlapping regions)
Adaptation & tweaks by jarredou
Original work AudioSR: Versatile Audio Super-resolution at Scale by Haohe Liu, Ke Chen, Qiao Tian, Wenwu Wang, Mark D. Plumbley
@article{liu2023audiosr,
title={{AudioSR}: Versatile Audio Super-resolution at Scale},
author={Liu, Haohe and Chen, Ke and Tian, Qiao and Wang, Wenwu and Plumbley, Mark D},
journal={arXiv preprint arXiv:2309.07314},
year={2023}
}