User Guide
Music & Audio
Generate and customize music and sound effects for your project
The Music tab lets you generate AI music that matches your project's mood. It is mixed into your video automatically — in the Render player and in every export.
Music Generation
Scene.video automatically generates a music prompt and tags from your script during analysis. The Music tab shows:
Prompt Ready State
When the music prompt is ready but not yet generated:
- Editable prompt — The AI-generated music prompt describing the desired soundtrack. You can edit this before generating.
- Tags — AI-generated genre/mood tags (e.g., "cinematic, ambient, dramatic")
- Model selector — Choose which music generation model to use
- Duration — Defaults to the total video duration. Shows a warning if the video duration exceeds the model's maximum.
- Generate Music button
Generating State
A spinner with "Generating music..." text. Music generation happens asynchronously.
Completed State
Once music is generated:
- Audio player — HTML5 audio controls for previewing the track
- Model selector — Switch models for regeneration
- Read-only prompt and tags — Shows what was used to generate
- Regenerate Music — Generate a new track with different settings
Failed State
If generation fails, the error message is displayed with a Retry button.
Music Models
| Model | Provider | Max Duration | Capabilities |
|---|---|---|---|
| ElevenLabs Music | ElevenLabs | 10 minutes | Prompt-based, instrumental (default) |
| MiniMax Music v2 | MiniMax | 5 minutes | Prompt + lyrics, instrumental |
| ACE-Step 1.5 | ACE Studio | 4 minutes | Open-source, prompt + lyrics |
| Lyria 2 | 30 seconds | Short-form, instrumental | |
| MMAudio V2 | MMAudio | 8 seconds | Video-to-audio SFX |
| ElevenLabs SFX | ElevenLabs | 22 seconds | Sound effects |
Duration Limits
Each model has a maximum duration. If your video exceeds it, a warning appears:
Video is 95s but lyria_2 max is 30s — music will be clamped.
How music reaches your video
There is no separate merge step. Once a track exists it is mixed in automatically wherever your project plays:
- In the Render player, in real time
- In any MP4 you export
Music is loudness-normalized so it sits under narration rather than fighting it, and a project-level volume control sets the final balance. You can also trim the track and choose where on the timeline it starts and stops, which is useful when a generated piece has a slow intro you would rather skip.
Turning music off in Render is instant and is remembered with the project — an export made with music off carries only scene and dialogue audio.
Sound effects per scene
Music is project-wide. For sound tied to a specific clip, use the scene's Post tab:
- Generate SFX watches the finished clip and replaces its audio with matched effects.
The video model can also produce its own ambient bed and foley, controlled by the Ambient & SFX field on the scene's Motion tab.