AI Music Video: Pair a Generated Song With Your Footage
Flow Music renders the song, the master and the stems. You bring the picture - here is the workflow that keeps a generated track usable in an edit.
A music video workflow fails for one of two reasons: the audio is not finished when the edit starts, or the audio cannot be licensed for the platform it is going to. A generated track removes the second problem, and this page covers the first.
Below: which export to use where, how to cut to an arrangement you specified yourself, and the crops and lengths each platform expects.
Where a music video workflow starts
Flow Music generates audio, not footage. That is a deliberate boundary: the studio gives you a finished, cleared track and the stems that go with it, and you decide what the picture is - your own footage, a lyric video, a slideshow, or an animated loop.
The four steps below are the workflow that keeps a generated track usable in an edit, whichever editor you already work in.
Finish the song first
Render the track and keep the take you want. A video built around a sketch means re-cutting every shot when the arrangement changes, so lock the audio first.
Export the right file
Download the WAV master for editing and the MP3 for quick previews. On Pro and Studio, also pull the stems so the vocal can be ducked under dialogue or a voice-over.
Cut to the arrangement
Use the sections you asked for in the brief as the edit points: the first chorus, the bridge, the final drop. Cutting on those turns makes a two-minute clip feel intentional.
Match the format
Render a 16:9 master for YouTube, a 9:16 cut for Shorts, Reels and TikTok, and a 1:1 crop for feed placements. Keep the hook inside the first three seconds.
Which export to use where
The studio exports a 44.1kHz master. Choose the file by what you are going to do with it, not by which one sounds marginally better in isolation.
| File | Format | Best for |
|---|---|---|
| Master, MP3 | 44.1kHz, 320kbps | Previews, social uploads, fast turnaround edits |
| Master, WAV | 44.1kHz, 16-bit | Any edit that will be re-encoded or mixed with dialogue |
| Stems, WAV | Vocals, drums, bass, other | Ducking the vocal under narration, remixing, extra fades |
| Extended edit | Up to 4 minutes | Longer cuts where the track has to cover a full scene |
Cutting to a generated arrangement
Generated tracks are predictable in a useful way: the sections you asked for are the sections you get. If the brief says the chorus arrives after one verse, that is where the picture should change.
Map the sections before you start cutting. Mark the intro, the first chorus, the bridge and the outro on the timeline, then place your strongest shot on the first chorus rather than at the start.
- Put the hook in the first three seconds for vertical formats
- Change the shot on the chorus rather than on the beat, so the edit survives a re-render
- Leave eight to twelve seconds of instrumental for a title card or an end slate
- Ask for an instrumental stem if a voice-over has to sit over the track
Platform lengths and crops
One render can cover several placements, but each one needs its own crop and its own in-point. Exporting a 16:9 master and cropping it later is faster than rendering the track again.
| Placement | Aspect and length | Note |
|---|---|---|
| YouTube | 16:9, full track | Keep the full arrangement; viewers expect an outro |
| Shorts, Reels, TikTok | 9:16, 15 to 60 seconds | Start at the chorus, end on the hook |
| Feed and carousel video | 1:1, under 30 seconds | Silent-first: the track should work without a lyric |
| Podcast and stream beds | 16:9 or audio only, 1:30 to 4:00 | Instrumental only, no vocal competing with the host |
Monetisation and claims
Paid plans include commercial rights for the tracks you generate, so a video using your own render can be monetised on the usual platforms. There is no sample to clear and no third-party recording inside the master, which is the usual source of a claim.
If you add third-party footage, music or a performance to the video, that material keeps its own terms - the studio covers the audio you generated, not what you edit on top of it.
Keep the brief with the export. If a platform ever asks how the audio was made, the brief plus the generation record is the answer.
What the studio does not render
There is no video renderer, no stock footage library and no automatic subtitle track. Nothing here claims to replace an editor.
What the studio does remove is the licensing question. A generated track with commercial rights and separated stems can be dropped into an edit immediately, which is usually the slowest part of putting music under footage.
AI music video questions, answered
Does Flow Music generate the video too?
No. The studio generates the song, the mixed master and the separated stems. You choose the picture - your own footage, a lyric video, a static visual or an animated loop - and cut it in whichever editor you already use.
What format are the exports?
A 44.1kHz WAV master and a 320kbps MP3, plus separated stems for vocals, drums, bass and the remaining instruments on Pro and Studio. The MP3 is for previews and quick uploads; the WAV is for anything that will be re-encoded.
How long should a video using a generated song be?
Full length for YouTube, where viewers expect the complete arrangement, and 15 to 60 seconds for vertical placements starting at the chorus. Render a 1:30 clip if you only need a short cut.
Can I use the track in a monetised video?
Yes. Paid plans include commercial rights for the tracks you generate, and there is no third-party recording inside the master to trigger a claim. Material you add to the edit keeps its own terms.
Can I dub a voice-over over the song?
Yes, and the stems make it straightforward: pull the instrumental stems up, drop or lower the vocal stem, and the narration sits cleanly over the arrangement instead of fighting it.
Do I need a different render for vertical video?
No. Render once at the length you need, then crop the picture for each placement. Only the in-point changes, which is why finishing the song before the edit saves the most time.
Can I generate background music specifically for footage?
Choose Instrumental only and describe the pacing in the brief - for example "steady electro bed, no build, 90 seconds of neutral energy". Instrumental renders cost the same as vocal renders of the same length.