I was recently wondering how well a language model trained on file formats like an Ableton Live would work with input prompts. I suppose we have all the pieces and only have to connect them as has been done in a growing number of areas.