1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
const axios = require('axios');
const api_key = "YOUR API-KEY";
const url = "https://api.segmind.com/v1/veed-lipsync";
const data = {
"video_url": "https://segmind-resources.s3.amazonaws.com/input/veed-lipsync-input-video-20260706.mp4",
"audio_url": "https://segmind-resources.s3.amazonaws.com/input/veed-lipsync-input-audio-20260706.mp3"
};
(async function() {
try {
const response = await axios.post(url, data, { headers: { 'x-api-key': api_key } });
console.log(response.data);
} catch (error) {
console.error('Error:', error.response.data);
}
})();Attributes
URL of the source video whose lip movements will be regenerated. Use a clear, front-facing clip with a well-lit, unobstructed face for best sync.
URL of the speech audio that drives the new lip movements. Any recorded voiceover or TTS track works; cleaner audio gives tighter sync.
To keep track of your credit usage, you can inspect the response headers of each API call. The x-remaining-credits property will indicate the number of remaining credits in your account. Ensure you monitor this value to avoid any disruptions in your API usage.
VEED Lipsync — AI Video Lip Sync Model
What is VEED Lipsync?
VEED Lipsync is a video-to-video model that re-syncs the lips of any talking-head video to a new speech audio track. Given a source video and an audio clip, it analyzes the on-screen face, then regenerates the mouth shapes and speech timing so the person appears to naturally speak the new audio — no manual keyframing, rotoscoping, or frame-by-frame animation required. Built by VEED and served through fal.ai, it is designed for realistic, production-grade lip synchronization across any aspect ratio.
The model works with existing footage, AI-generated avatars, and stock clips alike. Because it accepts any audio input, it slots directly into text-to-speech and dubbing pipelines: generate a voice track in a new language or with new dialogue, feed it alongside the original video, and get back a clean MP4 with matched lip motion.
Key Features
- Audio-driven lip resync onto any face in a source video, preserving the original identity and background.
- Format flexibility: video inputs in MP4, MOV, WebM, M4V, and GIF; audio in MP3, OGG, WAV, M4A, and AAC.
- Any aspect ratio — vertical, square, or widescreen all supported.
- TTS-compatible: works seamlessly with synthetic voice tracks for dubbing and localization.
- Simple two-input API: just
video_urlandaudio_url.
Best Use Cases
- Dubbing and localization — swap the audio for a translated voice track and re-sync lips for natural-looking localized content.
- Video rephrasing — change what a speaker says without re-shooting the scene.
- AI avatars and UGC ads — drive synthetic presenters or influencers with fresh scripts.
- Course and training content — update narration while keeping the on-camera talent.
Front-facing subjects with a clearly visible, well-lit face and minimal head movement produce the most convincing results.
Prompt Tips and Output Quality
VEED Lipsync has no text prompt — output quality is governed entirely by your two inputs. Use clean, high-quality speech audio with minimal background noise for the tightest sync. Choose source video where the face is unobstructed, sharp, and centered; heavy motion, occlusions, or profile angles reduce accuracy. Match the audio duration to the video length so the whole clip is covered. Expect roughly 2 to 2.5 minutes of processing per minute of input video.
FAQs
What does VEED Lipsync do? It re-syncs the lips in an existing video to a new audio track, making the person appear to speak the new speech naturally.
What inputs does it need? Two: a source video URL and a speech audio URL.
Can I use text-to-speech audio? Yes. Any audio works, including TTS output, which makes it ideal for dubbing and localization workflows.
What video and audio formats are supported? Video: MP4, MOV, WebM, M4V, GIF. Audio: MP3, OGG, WAV, M4A, AAC. Any aspect ratio is accepted.
How long does generation take? Approximately 2 to 2.5 minutes of processing per minute of input video.
What produces the best results? Clear, well-lit, front-facing faces with minimal head movement, paired with clean, high-quality audio.
Popular Models
