
Music for sleep
Long, slow and resolution-free. The goal is to give the brain nothing to predict, which makes melody a defect.
The same track rarely works in two places. A podcast bed and a TikTok loop have opposite requirements — one has to disappear under speech, the other has to grab in half a second. Each page below covers what that platform actually needs.
YouTube
Podcasts
TikTok
Twitch
Video Games
Instagram Reels
Commercials
Presentations
Sleep
Studying
Focus
Meditation
Relaxing
Workouts
Intros
Outros3–20 min
Background music you can monetise, without a Content ID claim.
10–30 s
A theme that survives two hundred episodes without wearing out.
15–60 s
Original sound that hooks in two seconds and is yours to use.
2–6 hours
Stream music that will not get your VOD muted.
60–180 s
Loops that survive being heard for six hours straight.
15–90 s
Reels audio that is cleared for a business account.
15 / 30 / 60 s
Advertising music cut to the slot, cleared for the use.
2–15 min
Music that supports a speaker instead of upstaging them.
30–120 min
Music engineered to stop holding your attention.
25–120 min
Music that supports concentration instead of competing with it.
45–120 min
Deliberately repetitive music for long stretches of deep work.
10–60 min
Sustained, pulse-free music that holds a session together.
20–90 min
Warm and slow, without tipping into sleep or ambience.
20–60 min
Music built around the tempo you are actually training at.
5–15 s
Five to fifteen seconds that set up everything after them.
10–30 s
The exit music that holds while people decide what to watch next.
Music fails on a platform for reasons that have little to do with whether it is any good. A podcast bed fails by filling the 300 Hz to 3 kHz band where speech lives, so the host becomes tiring to follow. A TikTok loop fails by having an intro, because the decision to keep watching happens before the intro ends. A stream background fails by having a memorable hook, which is charming for three minutes and intolerable by hour four. A commercial fails by running 31 seconds when the slot is 30.
None of those are quality problems, and none are fixable in the mix afterwards. They are decisions that have to be made before generation — length, dynamic range, where the hook sits, which frequencies to leave alone. That is what each page below covers: the constraints the platform actually imposes, and the prompt instructions that satisfy them.
Licensing is the second axis, and it varies more than people expect. Paid advertising is excluded from many stock licences. Monetised video invites automated claims even when your licence is valid. Music you generate and own removes that class of problem rather than managing it.
Royalty free music does not mean free music, and the gap between those two readings is where most licensing trouble starts. It means you pay once — or pay a subscription — and then owe no further per-play royalty. The recording still belongs to whoever made it. Your right to use it is a licence, and a licence has boundaries, an expiry, and a counterparty who can change the terms.
Those boundaries are what catch people. A licence covering an organic YouTube upload frequently excludes paid advertising. A subscription licence often covers work published while you subscribed and stops covering it when you stop paying. And because the same track is licensed to thousands of other creators, it is registered in Content ID by the rights holder — which is why a legitimately licensed track can still trigger a claim on your video and leave you disputing it.
Generated music sidesteps the category rather than negotiating it. On a plan that grants commercial rights, the track is yours: no per-play royalty, no expiry, no attribution line, no third party who can revise the terms later, and nothing already registered against your upload. The sixteen pages below cover what each destination needs on top of that — length, loudness, where the hook lands, and which parts of the spectrum to leave empty.
Inside
Each brief has its own tempo range, length, and way of failing.

Long, slow and resolution-free. The goal is to give the brain nothing to predict, which makes melody a defect.
1 / 5
Why this exists
A track is not good or bad in the abstract — it is right or wrong for the job.
Generate music for freePeople synchronise movement to a beat without meaning to, so BPM chosen from the activity beats BPM chosen from taste.
Language processing is a shared resource. That single fact explains why every study-music genre converges on instrumentals.
A build under a sentence steals attention from it. Flat dynamics are a feature in background work, not a failure.
A three-minute loop that fades out is worse than silence when the job is eight hours of sleep.
An instrumental bed generated to the length of your edit, mixed 12 to 18 dB below narration, on a plan granting commercial rights. Uniqueness is what removes the Content ID exposure that library tracks carry.
Royalty-free means no ongoing per-play fee, not free of charge or free of terms. What matters is whether your licence covers monetised and client work.
Sleep sits at 50 to 65 BPM, study at 60 to 90, focus around 100 to 125, and workouts at 120 to 150 depending on the activity. Tempo does more of the work than genre.
Yes, under the commercial terms of your plan, which are shown in the app before you generate. Downloads carry no watermark, including on the free tier.
As long as the thing it sits under. Generating to the exact length lets the track resolve, where a fade mid-phrase reads as an edit to a listener even if they cannot say why.
Rarely. Lyrics compete with narration and with reading, which is why almost every brief on this page specifies instrumental.
Generate something built for where it will actually play.