Open JamesDuncanRobertson opened 1 year ago
Thank you for this great example. The released pretrained model is only trained on recordings that does not have this many reverb. I will look into it later.
It has happened on every recording I tried with.
The outputted MIDI files often simulates reverb in the audio recording by replaying multiple notes rapidly. Are the developers familiar with this? Any fixes?