Multi-speaker and multi-angle lip sync

What multi-person support means

Bonframe supports 15+ faces and multi-angle footage. For interviews and group conversations, the important review is whether each voice belongs to the correct visible person in each scene.

Face count is not an unconditional guarantee for every shot. Overlapping speech, heavy occlusion, small faces, fast motion and poor source quality can affect the result. Review your actual material rather than assuming a large-cast demo covers every scene.

Prepare a lip-sync task

1. Confirm the translated dialogue and speech track before generating lip sync.
2. Check the correspondence between speakers, voices and faces, especially when a person reappears after a cut.
3. Choose standard or HD lip sync from the options available to your license and prepare the matching local model.
4. Test a short passage containing your typical camera changes, profiles or head turns.
5. Review the result with sound and inspect returning faces, cut points, timing and visible mouths before processing the full video.

Standard and HD

Standard and HD use different model options and license entitlements. Pro includes the HD lip-sync model; select it only when the required model and device resources are ready. Higher quality can require more processing time and memory.

Do not change text, voice and visual settings at the same time when diagnosing a problem. Keep a reviewed audio track so that visual comparisons are meaningful.

Review difficult scenes separately

• Side profiles and head turns: inspect the transitions, not only the front-facing frames.
• Microphones, hands and other occlusions: check whether visible mouth changes remain consistent.
• Several people on screen: verify that a listening person does not receive the speaking person’s mouth movement.
• Camera cuts: inspect the frames immediately before and after the cut.

Related guides

• Original-voice dubbing and voice review
• Devices and local model setup
• Review and export your result