7 Best Wav2Lip Alternatives in 2026: Open-Source and Commercial Lip Sync

Wav2Lip (ACM Multimedia 2020) made it practical to lip sync any face in a video to new audio, and it's still one of the most forked lip-sync projects. Its README now opens with Sync Labs' commercial API, says the open-source version is for personal, research and non-commercial use, and adds that, because the models were trained on LRS2, any commercial use is strictly prohibited. Newer models are sharper, and several are licensed for commercial use.

Commercial use

Wav2Lip's README restricts the open-source release to non-commercial use and points businesses to Sync Labs.

Sharpness

Wav2Lip's mouth region is often blurry; newer models such as LatentSync 1.6 train at 512×512.

Maintenance

The repository has no releases, and its recent commits are README changes.

Live use

Wav2Lip processes recorded video; it isn't built for a live, streaming avatar.

Licence

Check both the code and the model weights: several projects bundle third-party models with their own terms.

Input

Video + audio (dubbing), image + audio (talking head), or audio alone driving a live avatar.

Hardware

Diffusion models need large GPUs; hosted APIs and in-browser SDKs need none.

Pricing model and licence

Per minute, per hour, per monthly active user, or free with a licence: model your real conversation volume.

1. 1. Best hosted lip-sync API for video

   Sync Labs (sync.so)

   The commercial lip-sync API from the team behind Wav2Lip, with newer models (lipsync-2, lipsync-2-pro, sync-3) that take a video and audio and return a lip-synced video.

   Pricing: Plans $5–249/month plus $0.02–0.167 per second of video by model (about $1.20–10 a minute).

   Best for: Commercial dubbing and video localisation without running GPUs.

2. 2. Best open-source quality for offline dubbing

   LatentSync

   ByteDance's lip-sync model built on audio-conditioned latent diffusion. Version 1.6 is trained at 512×512 for sharper mouths and needs about 18 GB of GPU memory to run.

   Pricing: Free (Apache-2.0); hosted versions on Hugging Face and Replicate.

   Best for: Rendering dubbed or re-voiced video where quality matters more than speed.

3. 3. Best open-source real-time lip sync for video

   MuseTalk

   Open-source model that re-renders the mouth region of a face video from audio, in latent space. Version 1.5 runs at "30fps+ on an NVIDIA Tesla V100", with Chinese, English and Japanese audio.

   Pricing: Free: code MIT, model weights usable commercially (bundled third-party models keep their own licences).

   Best for: Teams with a GPU who want photoreal lip sync on their own servers.

4. 4. Best for a talking head from one photo

   SadTalker

   Turns a single image into a talking-head video by predicting 3D head motion from audio (CVPR 2023). Its code hasn't been updated since 2023.

   Pricing: Free (Apache-2.0, third-party parts excepted).

   Best for: Animating a still portrait offline.

5. 5. Best price per minute for your own voice agentOur product

   Talking Avatar

   Talking Avatar Realistic Avatars is a real-time avatar API: pick a ready-made photorealistic avatar and it speaks your voice agent's audio (OpenAI Realtime, LiveKit, Pipecat or any TTS) with lip sync, at 832×468, 25 fps, on your page over WebRTC or in your LiveKit room. The MIT-licensed open-source SDK lip syncs your own 3D avatars in the user's browser.

   Why it fits here

   • Talking Avatar Open Source: a 1 MB model that lip syncs your own 3D avatar in the browser from any audio, live
   • Talking Avatar Realistic Avatars: ready-made photorealistic avatars at 832×468, 25 fps, for $0.02 a minute
   • Built for voice agents: OpenAI Realtime, LiveKit, Pipecat or any TTS
   • Commercial use included on both: the open-source SDK is MIT

   Pros

   • Live, streaming lip sync rather than batch video processing
   • No GPU: Open Source runs on the user's device
   • Free for commercial use (MIT), or a published per-minute price for realistic avatars

   Cons

   • Doesn't re-voice recorded video, which is what most Wav2Lip users do
   • Talking Avatar Open Source animates 3D .glb avatars, not real footage
   • Talking Avatar Realistic Avatars uses ready-made avatars only, not your own face or footage

   Pricing: Talking Avatar Realistic Avatars $0.02/min, billed per second, no subscription; open-source SDK free (MIT).

   Best for: Developers who already run a voice agent and want a face for it at the lowest per-minute price.

6. 6. Best for fast portrait animation from a driving video

   LivePortrait

   Animates a portrait from a driving video (people, cats and dogs) at 12.8 ms a frame on an RTX 4090. It's driven by video, not audio, so lip sync needs another model.

   Pricing: Free: code MIT; its InsightFace models are non-commercial and must be swapped for commercial use.

   Best for: Expression and head-motion transfer from a performer.

7. 7. Best research quality for long portrait videos

   Hallo (Hallo2, Hallo3)

   Fudan's diffusion models that animate a portrait from audio: Hallo2 for long, high-resolution video and Hallo3 (on CogVideoX-5B) for dynamic scenes. Tested on A100 and H100 GPUs.

   Pricing: Free: MIT, but CodeFormer (Hallo2) and CogVideoX (Hallo3) licence terms apply.

   Best for: Research and high-end offline renders.

Start from what you need to produce, then check the licence for both code and weights.

1. Re-voicing recorded video commercially: try Sync Labs, or self-host LatentSync or MuseTalk.
2. A talking head from one photo: SadTalker or Hallo, offline.
3. A live avatar for a voice agent: Talking Avatar, with Open Source in the browser or Realistic Avatars streamed to your page.
4. Read the licence of every bundled model (face detectors, VAEs, Whisper) before shipping.

• Commercial video dubbing without GPUs: Sync Labs.
• Open-source, commercially usable, best quality: LatentSync.
• Open-source and real time on a GPU: MuseTalk.
• A talking head from a single photo: SadTalker, or Hallo for higher quality.
• A live avatar for a voice agent: Talking Avatar (Open Source, or Talking Avatar Realistic Avatars).
• Transfer a performer's expressions: LivePortrait.