Skip to main content

VEED Lipsync 2.0 re-syncs the mouth of one speaker in existing footage to a new audio track, keeping the video's resolution. No speaker training. Billed per second of audio.

From $0.084/s

Workbench

Input

Lip sync
1/1

Drop a video here, or browse · 1 file · Max 90 MB · MP4 · WEBM

  • 1
1 / 1 files
Required. MP4 or WebM of one speaker with a visible face. The output keeps its frame size. at least 1 characters
1/1
1 / 1 files
Required. The new speech, up to five minutes. Its length sets the length of the result and the charge. at least 1 characters
Safety checkerOff

SpicyAPI does not run a safety checker on this model. Your prompt and media go to the model as written, and the model itself decides what comes back.

Checking your account before enabling submission and pricing.

Output

Ready for a new run
Example result

Related models and next tasks

Missing the model you need?

NEW AUDIO, SAME TAKE

Give finished footage a new line and the speaker's mouth follows it.

Key numbers

Speakers
1
One visible speaker per clip; group scenes are not supported.
Training
None
No sample of the speaker is needed, the clip itself is enough.
Length
Audio
The result is as long as the new audio, shorter or longer than the footage.

Before you spend

Lip sync

One speaker per clip. Audio up to five minutes. The face should stay visible and fairly still; fast movement, profiles and covered mouths reduce accuracy. The model does not translate or write text: bring the finished audio yourself.

What it does

Same take, new words

A balcony shot re-voiced with a five-second line

We gave Lipsync 2.0 a six-second clip of a man in a white linen shirt on a hotel balcony and a new recorded line. His mouth now forms the new words while the city lights, the shirt and his half smile stay as filmed. The result was 864×496 and 5.08 seconds long, trimmed to the audio.

Before and after

The same moment, a different mouth shape

Two stills from the same moment of the balcony clip: the original take on the left with the man smiling, closed-mouthed, and the VEED Lipsync 2.0 result on the right with his mouth open mid-word.

Two seconds in, side by side: on the left the original take with his lips closed in a smile, on the right the result with his mouth open on the new line. The shirt, the railing and the lights behind him are unchanged.

No speaker training

The clip is all the reference it needs

There is no voice sample, face upload or fine-tuning step. Swap the audio file and the same footage carries a different line.

Where it fits

  • Fix a line after the shoot

    Change a word, a price or a product name in a finished take without bringing the speaker back to the set.

  • Translated voice-over that looks native

    Put a translated voice track on the original footage and let the mouth follow the new language.

  • New dialogue for generated clips

    A character from a video model can say a different line than the one it was generated with.

Model reference

VEED Lipsync 2.0 API reference

VEED Lipsync 2.0 API specifications

Model
VEED Lipsync 2.0
Required inputs
Video of one speaker (video_url) and new speech (audio_url)
Resolution
Same as the input video
Length
Follows the audio, up to 5 minutes
Output
MP4 with the new audio, 25 fps
Billing
Per second of audio, rounded up

How to call the VEED Lipsync 2.0 API

  1. Pick a take with a clear face

    Front-facing, well-lit and steady works best. Fast head turns, hands over the mouth and hair across the lips make the result less precise.

  2. Prepare the new line

    Record it, generate it with a voice model or take a translated voice track. One voice, no music: noisy audio is best cleaned first.

  3. Send both links

    Pass the footage in video_url and the speech in audio_url. The finished clip comes back at the resolution of the footage.

VEED Lipsync 2.0 API questions