AI music for YouTube: a 120 second ceiling changes the plan

You type a description in your own words and a finished track comes back. Supply lyrics and it is sung. Leave the box empty and it is instrumental.

Lengths are 15, 30, 60, 90 and 120 seconds. The price is a flat 20 credits at every one of them.

For a pre-roll ad or a short, that range covers everything you need.

For a twelve minute video, it does not. Two minutes is the longest piece, and stitching several together is a job for whatever editor you cut the video in.

So the strong use here is ad music and intro music rather than a bed for a long-form upload.

That is not a small category. A channel running pre-roll, shorts and an intro needs three or four tracks a quarter, and every one of them fits inside the ceiling.

Anybody arriving looking for forty minutes of background for a study video should stop reading now and buy a library subscription.

An audio matching system compares your upload against a database of registered recordings.

A generated track is made for you rather than licensed from a catalogue, so there is no identical recording in circulation for it to find.

That removes the most common annoyance on the platform, which is a claim appearing on a video using music somebody genuinely paid for.

It is not a guarantee and nobody should sell it as one. Matching systems change. A claim can be disputed either way.

Confirm the usage terms that apply to your own account before you build a channel identity on a track.

Keep the description you typed, the date and the file. That record is what answers a dispute. It costs ten seconds to save.

This platform splits across phone speakers, laptop speakers, television soundbars and headphones. Each one reproduces a different part of your track.

Phone and laptop speakers give you almost nothing below roughly 500 Hz, which is where most of the weight in a track sits.

That means the midrange carries the whole piece for a large share of your audience, and the midrange is where speech lives.

So a track described as full and warm competes with your own narration on exactly the devices most people are using.

Describe space rather than size. Sparse, mostly high percussion, almost nothing in the middle is a better brief than anything with big in it.

Then check the video on a phone at half volume, which is the realistic condition rather than the flattering one.

A sung vocal and a spoken narration compete for the same frequencies. They compete for the same attention too. A listener follows neither.

Use lyrics on the intro where nobody is talking, and instrumental everywhere the narration runs.

That is two tracks at 20 credits each, which is 40 credits and a fix no level adjustment can match.

In the ad editor the music controls are turn it on, pick the track, and set how loud it sits. No ducking, no stems, no tempo slider.

If you need automatic ducking under narration, that is a full audio editor's job and it is not what this module does.

Choosing which shape the piece is before you generate saves the 20 credits you would otherwise spend twice.

The intro is where the economics here beat everything else. One track at 20 credits sits in front of every video you publish for a year.

Nothing in a subscription library does that more cheaply, and nothing in a library is yours alone.

So spend five attempts and 100 credits getting it right rather than accepting the first return.

Save the description you settled on. Regenerating from the same words gives you a sibling track when the intro needs refreshing.

The same logic covers pre-roll ads, shorts and the 9:16 ads you run alongside the channel.

Four plain limits sit around that. Two minutes is the ceiling on a single track, so a long video needs several pieces joined outside this product.

For weekly long-form uploads, a subscription library is the cheaper and more practical answer to background.

There is no tempo slider, no stem export and no way to edit a track after it comes back. You generate again for 20 credits instead.

And a specific recognisable song is a negotiation with a rights holder rather than a generation problem.