What this AI music video generator creates
AIRAP creates a short performance featuring the people in your portraits. It builds a scene image, then generates movement and optional original sound. The output is a clip you can download as MP4, rather than a complete edited music video with many scenes.
Decide what the finished piece needs to do before choosing a scene. A personal dedication may work as one short performance; a release-length video usually needs several clips, an existing recording and an editor. AIRAP supplies generated performance clips, so plan the final sequence around that scope.
Photos guide the people, the theme supplies the setting, and the optional topic guides original sound. A detailed music brief cannot repair an unclear face or turn one request into a multi-scene production.
Choose a solo session or a two-person performance
For a keyboard-led live-session look, choose Tiny Desk and upload one portrait. The selected 10-, 12- or 15-second arrangement keeps your reference subject as the lead, with a backing ensemble in the room. Hotel Lobby and On The Radar use two portraits for a rap duet. Halloween uses one portrait for a 10-second solo freestyle.
Prepare portraits before planning a sequence. Face the camera, keep eyes and mouth visible, use soft frontal light and leave space around the head. A dark concert photo may suggest the right mood while providing too little facial detail.
- Check the actual file, not just its filename
- Use JPEG, PNG or WebP up to 2 MB per photo. Export HEIC or another unsupported format before uploading; changing its extension does not convert it. An image can also upload successfully while still being blurry or too small to describe the face well.
- Prepare a pair as two individual references
- Hotel Lobby and On The Radar require two separate portrait uploads. Give each person a clear image with a similar level of detail. Avoid group selfies, overlapping faces and heavy filters; a matching original background is unnecessary because the performance scene is generated.
- Keep a solo reference unambiguous
- Tiny Desk and Halloween use one portrait each. Crop other people out without cutting into the lead person's head. The Tiny Desk backing ensemble is generated scenery and supporting cast; you do not upload portraits for those musicians.
Match the image to the music idea
An intimate dedication fits a close live-session scene; an energetic exchange fits a duo at the microphone. Pick the visual theme first. Describe the music subject and mood in the audio topic without using the field to request a different set, camera ratio or model.
Hotel Lobby offers four stages for the same two-person format. These are visual choices, not four separate song-upload modes. Compare the background, light and microphone arrangement with the mood of your planned piece.
- Orange studio: a graphic, direct performance
- The orange backdrop and hanging microphone suit a playful duet or bold short-form introduction. The simple set keeps the performers prominent when the video is viewed on a small screen.
- Luxury Lobby: a celebration setting
- Warm chandeliers, cream marble and two vintage microphones suit a milestone or formal dedication. Choose this when the setting should carry a sense of occasion alongside the vocal message.
- Studio Booth: a close recording-session look
- Dark acoustic panels, violet-blue accents and a shared microphone support a personal or collaborative mood. It is a useful visual match for a message focused on the relationship between two performers.
- Street Cypher: an outdoor duet
- A night alley, graffiti textures and separate microphones suit an energetic exchange. The stage retains two leads; asking for a crowd or a car chase in the audio field does not redesign it.
Working with an existing song
The public generator currently creates original audio from a topic and does not accept a song upload. If you already have a recording, you can create a silent visual clip and combine it with your track in an external editor. Check your rights to that recording. Replacing audio after download does not guarantee the lips or gestures will synchronize with it.
If your recording is finished, disable generated audio and use the downloaded clip as visual material in an editor. Arrange cuts against the track there. A song URL or full lyric sheet in the topic field does not import a recording.
For original audio, specify an occasion, one detail and a preferred language. For an existing recording, inspect visible mouth movement before pairing it with the track; visual atmosphere and exact lyric synchronization are different requirements.
Plan a longer music video as separate clips
Choose a consistent ratio for every shot. Plan each short clip around one clear moment, then download and edit the clips together. Keep the same source portrait to support continuity, while expecting some variation across independent generations. Review the cost of each request before confirming.
For a 30-second idea, sketch three ten-second moments: introduce the people, develop the story, then close on the message. Hotel Lobby or On The Radar can supply short duet clips. Tiny Desk offers 10-, 12- or 15-second sessions to fit the plan. Separate generations do not guarantee one continuous take.
Keep portraits, theme, stage, model and ratio consistent where continuity matters. Independent generations can still change clothing, expressions or backgrounds. Inspect the first result before ordering the rest, then join usable shots where changes feel deliberate.
Leave room for titles, transitions and an ending in your editor. Each generated clip has its own quote, so budget for the number of usable shots needed.
Set the format for where you will share
Choose the format for where you will share: a vertical phone clip, a landscape frame or a square post. The settings show supported aspect ratios and durations; some themes use a fixed length.
Choose 9:16 for a phone-first vertical composition, 16:9 for a landscape sequence or 1:1 for a square post. Check where faces and microphones sit before adding captions in an editor. Cropping landscape footage into a vertical frame later can remove one person from a duet.
Hotel Lobby and On The Radar offer 5, 8, 10, 12 or 15 seconds. Tiny Desk offers 10, 12 or 15 seconds and Halloween is fixed at 10. Assign one achievable moment to the selected theme's duration.
- 5 seconds: an insert or opening beat
- Useful for one compact exchange or a brief visual accent in an edit. Leave a complicated story for a longer clip rather than squeezing it into the topic.
- 8 seconds: a short standalone idea
- Allows a little context and a response while keeping the performance compact. Plan one emotional point, especially if the finished post also needs a caption or title.
- 10 seconds: a versatile sequence building block
- A practical choice for planning several short moments. It can carry an occasion and a personal detail without making each request responsible for an entire music video.
- 12 seconds: more breathing room
- Consider it when an exchange needs a setup and a closing thought. Generated phrasing can vary, so treat the duration as a container rather than an exact lyric timetable.
- 15 seconds: a fuller performance moment
- Use it for a longer duet or a 15-second Tiny Desk session. Keep the brief focused; extra seconds do not guarantee more scene changes or a complete song structure.
Review the model and credit quote
Seedance 2.0 Mini and Fast offer 480p and 720p. Seedance 2.0 and 2.5 also offer 1080p. Start with the settings you actually need: the exact credit quote updates with model, resolution and duration. Native audio is included. Review the final quote before you confirm generation.
Choose a resolution for the delivery size and check the exact model/duration quote. More pixels cannot repair a weak reference. Review faces and movement before generating multiple clips with the same settings.
Before you confirm generation
Check the people and theme, read the audio topic, confirm the ratio and resolution and review the displayed credits. Uploading photos alone does not start a video job. Once generation is confirmed, follow the saved task before submitting another request.
- A weak likeness across clips
- Use the clearest consistent portrait and keep visual choices stable. Watch for changes during movement before building an edit around the clip. Independent retries may introduce different variations.
- The mood, wording or language misses the brief
- Remove competing requests; name one occasion, detail and language. Keep technical settings in their controls. Select a different location through the theme or stage rather than the topic.
- The clip does not fit the final edit
- Check ratio, face placement and duration against the storyboard. Add overlays and the final recording in an external editor. Replacement audio does not retime generated mouths or gestures.
- The request is still processing or needs attention
- Check My Videos and the task's status and credit/refund information before generating again. Creative quality is a separate assessment. Each fresh request has its own quote and charge.