I think we still need the step where the AI learns what a high quality sound library sounds like and then applies the previously learned abilities by triggering sounds of that library via MIDI.
That way you'd get perfect audio quality with the creativity of a musical AI.
That way you'd get perfect audio quality with the creativity of a musical AI.