Original work Apollo: Band-sequence Modeling for High-Quality Music Restoration in Compressed Audio
Apollo audio restoration Colab fork
Original model was trained to restore/enhance lossy mp3 audio with bitrate <= 128 kbps.
changelog:
v0.4
- added: config loader
- added: lew's separated vocals enhancer v2 beta
v0.3
- lew's separated vocals enhancer model added
v0.2
- added overlap feature
- new inference.py created for easier local CLI use
v0.1
- added chunking for long audio inputs
- added "dual mono" processing for stereo audio input (processing each channel independently)