
What AI music actually is
How text-to-music models work, what they can and cannot do, and where the technology is genuinely useful.
Guides
How to make music with AI, from the first prompt to a finished, licensed track. These guides start at zero — no DAW, no theory, no gear — and go as far as monetisation, Content ID and what the copyright position actually is. Written plainly, with the trade-offs left in.
In short
Start here
How-to
Inside
Fundamentals first, then the how-to, then putting it to work.

How text-to-music models work, what they can and cannot do, and where the technology is genuinely useful.
1 / 5
Every technique has a cost. A guide that only lists benefits is advertising with headings.
Numbers, tempos, frequencies and named instruments — because "add warmth" is not an instruction anyone can follow.
Drafted with AI assistance and reviewed by the team, with corrections made in place and dated.
Where a figure appears, it is sourced. Where we measured something ourselves, the method is published.
Describe a track — genre, tempo in BPM, instruments, mood and whether it has vocals — then generate, listen and change one thing. Most people reach something usable in three or four attempts.
No, but musical vocabulary helps enormously. Knowing you want 90 BPM rather than "medium speed" is the difference between getting what you imagined and getting an average of everything.
Naming six things: genre, tempo, key or mood, instrumentation, vocal type and production era. A prompt that specifies five concrete things beats a poetic sentence that specifies none.
Generally yes, under the terms of the platform that generated it. What varies is whether commercial use is included on your tier and whether the output is watermarked.
Yes, if your plan grants commercial rights and the download has no watermark or attribution requirement. Those two conditions decide it, not the fact that a model made the audio.
A single generation takes about a minute, and reaching a track you actually want usually takes three or four. Budget five to ten minutes rather than an afternoon.
To make music with AI you describe the track in words — genre, tempo in BPM, instruments, mood, and whether it has vocals — then generate, listen and correct the description. No instrument, DAW or music theory is required, and the loop takes about a minute per attempt. These guides cover that method in detail, then the parts that come after it: writing prompts that work, choosing between generators, clearing the track for commercial use, and monetising video that contains it.