The future is a little scary, but I don’t think it involves DAWS.
So, I wrote a little piece called “The Venetian” in Dorico. PDF attached (engraving is not great, it’s just for internal consumption).
Then, I exported a basic playback. And imported into Suno. I let it describe the music and create the prompt itself (which it does by default).
And this is what came out! It’s almost perfect, note for note, plus a few additions. It completely screws up the middle few bars, though.
But frankly, it brings out the very best that is to be hoped for in this little composition.
Also, because it is 85% faithful to the original, I still have the satisfaction of calling it my own composition. (And probably legally, as well, can copyright this.)
If you are in the US, look up “Circular 1” from the copyright office. I believe that - by virtue of posting it here as the pdf and dorico audio (which may be more faithful to that copy) - the work has been “published”! US law protects a piece when it’s created. Proving it’s yours is another matter which is what copyright registration does for you. (It does not prove a work is “original”, however!) In the circular, there is a section on publication with some of the rights conferred (and possible responsibilities). You can and should add the (c) date if you’re in the US!
If Dorico could have an AI-playback layer, we could achieve a fully realistic orchestral playback. With the AI given both audio and notation data as input.
However, the state-of-the-art right now is that the composer has got to be prepared to relinquish quite a lot of creative control. It’s like writing a piece of music and handing it off to an arranger with instructions.
Still, if, like me, you prefer focusing on the compositional process rather than the playback, things are moving in an interesting direction!
Wow… Pete… thanks for posting..! Suno is very clever stuff…
Also, can you now go in and edit (with prompts) that existing output, to change between bars xx - xx for example.? Change instrumentation.? Alter the tempo rall at end.? etc…
From your ‘bare bones’ idea, to turn this out - the sound reproduction and the setting of the piece, coupled with the arrangement, sensitivity/dynamics throughout, is impressive…
Wrestling with Suno is like sumo wrestling – it’s a thankless, very difficult task. There is some sort of editor, but if you ask it to change one bit of the piece, it has to redo the entire piece (in my very limited experience) so it takes a minute or two, and you have no idea what the result will be like.
There is very little fine-grained control, as far as I can tell.
Hopefulll with time this will change, but it feels like it’s a limitation of AI. AI isn’t a tool in the classic sense.
What I’m hoping for is that there exists an AI algorithm in Dorico capable of writing expression data, which can then be assigned to an instrument. Perhaps even a second line for vibrato. Since I only use performance based instruments—like Audio, Modeling, or Aaron Venture—I already have very homogeneous playbacks. The only issue so far has been the long notes. This way, I’d have the ability to define details within the instruments themselves, make precise adjustments, and still play back my music without having to re-render the piece.
It totally butchers the opening phrase, playing your slurred notes staccato, and adds a C on the end of the phrase. On the second phrase, it adds additional notes which completely change the resolution. That’s … kinda terrible and if I performed that poorly no one would ever hire me. I honestly didn’t get much further than that.
I don’t consider this a “performance” of the written piece. As I just mentioned above, it’s a “cover” of the original. But what impressed me was how close to the original it is. Sufficiently so that I think it could definitely be copyrighted, for instance.
I guess I’m failing to see what the point is then. I’m all for using AI with Dorico, and have had some limited success with having Claude write Expression and Percussion Maps already. I would be interested in exploring some AI playback options for mockups, but from what you’ve demonstrated, Suno can’t play back what you wrote in Dorico, so seems sort of useless. It certainly isn’t a replacement for Dorico’s playback or a DAW.
I have no interest in having AI rearrange or “cover” my music. Composition and arranging is the fun part! I want AI to handle the mundane “busy work” aspects of Dorico work, not the creative parts.
