AI-assisted music, video and finishing tools; completion varies by project.
Tofu and the other characters give the music a world to inhabit. Suno helps generate musical material, but selection, sequencing, artwork and the decision to keep or discard a track still shape the result. A folder full of songs is not automatically an album.
I built suno-master to make recurring finishing problems easier to inspect: loudness differences, stereo compatibility and a harsh or muddy mix. The reports give me measurements to compare with what I hear. They do not get the final vote.
One useful correction was to make tonal adjustments optional. A sharp peak in a full-mix spectrum can be a played note rather than a defect. Automatically cutting every suspicious peak can remove the body of the music. The tool’s documented approach now keeps those suggestions advisory unless I choose to apply them.
That is the part of AI music I find most interesting: knowing when to stop automating. I can measure a track and still dislike it. I can prefer something imperfect because it serves the song. The SaltScribe creative and distribution experiments extend that work into how the music is presented, without treating a planned campaign as a proven audience result.