AI music you can
open and change.

A generated audio file is the end of the process: the parts are mixed together, and changing the bass line or the chords means generating again and hoping. A generated project keeps the decisions apart — form, parts, notes, instruments, mix — so you can change one without losing the others.

Neither is “better” in general: an audio file is quick to use as it is; a project is what you want when the idea is a starting point.

• Form — the song is a sequence of sections (intro, theme, breakdown, return…) you can move, vary or extend.
• Parts and notes — every part is its own track with its own notes or drum steps.
• Harmony — the chords are one shared context; change a chord and the parts that follow it know.
• Sound — every part plays on an instrument whose settings you can open and change.
• Mix — levels, panning, effects, automation and a master you can switch or fit.

HTC splits the work the way a production does. It reads your request, chooses a style’s grammar, plans the form, composes the parts together around one musical idea, performs them, picks instruments for each role and mixes the result. Each step is a separate, versioned part of the project — which is why you can change one without regenerating the rest.

A song also remembers which model made it. When HTC improves, your existing projects stay exactly as they were.

• Sketch fast — one request, six directions, keep the one that speaks to you.
• Build by layers — start from drums and bass and let HTC propose the next part, take by take.
• Learn arrangement — open a song and see how its sections are built and why it lifts where it lifts.
• Finish elsewhere — export stems and continue in your usual software.

HTC does not sing, does not clone artists and does not do orchestral or live-band music. It is a beta: results vary, and some are better than others. The point is that every result is a place to start from.