Turn a song and portrait into a singer video

Open Tools → Singer Video Generator, or choose Generate Singer Video from a completed song’s menu. The tool animates a singer image against an audio clip; it is different from the visualizer available in the normal download dialog.

Choose the two source assets

For the singer, select a preset or upload a JPG, PNG, or WebP portrait that you are allowed to use. A clear, front-facing face with even lighting is easier to animate than a distant subject, heavy obstruction, or several people.

For audio, choose Upload Audio or Song Library. Supported uploads include common formats such as MP3, WAV, M4A, AAC, and OGG. A completed Library song avoids a second upload.

The generated clip currently supports 5–120 seconds. If the selected audio is longer than the limit for your plan, use the trim step to choose the section that matters. Include a short lead-in before the first vocal so the motion does not begin abruptly.

Select a frame for the destination

Choose an aspect ratio for the channel where the video will be used. Vertical and square frames suit short-form feeds; landscape suits standard video pages. Some non-original ratios require paid access.

The page calculates credits from the actual clip duration: the current rule is 10 credits for each started 5-second block. Review the displayed estimate before creating the task.

Generate, monitor, and review

Choose Create Singer Video. Recent tasks stay on the page and move through queued, generating, completed, or failed states. You can continue preparing another render while a task is processing.

When complete, review lip timing, facial stability, framing, and the transition at the clip boundaries. Open the detail page to share or download when your plan allows it. If the account displays an expiring-storage notice, save the final file promptly rather than treating the gallery as permanent archive storage.

Keep the original portrait, audio source, selected time range, and final download together so a later revision can reproduce the same setup.

For revisions, change one variable at a time. Keep the same clip while testing a new portrait, or keep the portrait while adjusting the crop. This makes credit use easier to evaluate and reveals which change actually improved the result.

Use the final revised song as the video source

Open /ai-music-video-generator and choose the completed version from the song library, or upload its downloaded audio. A cover, refined result or song with newly added vocals should be reviewed before rendering the portrait video. Trim a coherent 5–120 second passage within your plan’s limit and recheck the portrait rights, framing and cost. The displayed estimate uses 10 credits per started 5-second block. Keep the same audio excerpt while comparing portraits so you can identify what improved. Save the final file after checking lip timing and the ending.