What is the hardest aspect of blind automatic music transcription?

Answer 1

I'm actually a Data Scientist by trade, so if I were to hazard a guess it would be somewhere around the combination of models to detect unique instruments and designing the audio processing model to accurately transcribe each specific instrument. Not to mention beat detection or dynamics, the latter of which is inherently subjective. I'd say it's ALL pretty complicated and difficult, but there are some more difficult areas than others. Hope that helps!