So Fable can act by itself as audio to midi decoder now? No need for specialised models, it just listens like a person and plays it?