Render a visualizer after the Suno track is the one you will post

create_music_video turns an in-app Suno generate into a simple waveform/visualizer clip — the fastest path from audio-only to something postable. It needs taskId and audioId from a prior generate_music. It is not Movie mode. It is not a scene-by-scene narrative. It is a finishing render of a song you are keeping. A discarded generate does not need a visualizer.

Create a music video from a song is the job. Billed per render, separate from the track. estimate_cost before you click through. For a real picture, use the AI video generator around the finished track — still after the song is the song.

Audio lock, then pixels

Track lock here is: this audioId is the one you will post, at this length, with these lyrics if any. If you are still extending, covering, or rewriting, you do not have a parent. The visualizer will be of the short version, or the wrong arrangement.

People render the visualizer first because they want "something to look at" while they decide if they like the song. Listen to the song. The visualizer is how you publish a decision, not how you make one.

This is a later step relative to the video pipeline too. If the actual deliverable is a brand film with a bed, you wanted attach_audio_to_video on locked picture, not a waveform of the bed. create_music_video is for when the song is the product.

What you get

A straightforward visualizer. No shot list. If you needed a story, that is a different generate, scored with this track after both are keepers. Do not run this tool as a "placeholder film." You will get attached to a waveform and still owe a film.

The AI music generator is the parent surface. Keep the ids. Losing them is how you regenerate the song to render a video, which is the leak with extra steps.

Order

1. Generate (and extend) until the track is the release.
2. Render create_music_video once.
3. If you instead needed picture, generate picture to the runtime and mix. Skip the visualizer.

FAQ

Can I point this at an uploaded MP3?

No. In-app generate_music ids only. An upload is a different editor job.

Should I split stems before the visualizer?

Only if the visualizer should be of a stem, and only after that stem is the release. Usually you visualize the full mix.

Is this cheaper than generating a real music video?

It is a different object. A waveform is not a scene. Do not use "cheap visualizer" as a first step toward a film. You will still make the film, plus this.

What if I hate the visualizer but like the song?

Keep the song. Do not regenerate the audio to reroll the waveform unless you also hate the audio. Rerender the video tool if it allows; do not spend a new generate_music as a video retry.