---
title: Lip sync video and make talking avatars
display: '*Lip sync* video and make talking avatars'
thumb: video/lip-sync-and-talking-avatars/thumb.webp
nav_title: Lip sync and avatars
summary:
  - "Re-sync the mouth of an existing clip to new speech, or animate one portrait into a talking presenter."
  - "Clean, natural speech and a well-lit, visible mouth decide most of the result: fix inputs before models."
  - "Find both kinds of model in the model library of Create > Videos; runs are billed by length, so trim the source first."
description: Sync new speech to an existing clip, or animate a portrait into a presenter from an audio track, and check mouth shapes and identity.
section: video
order: 80
type: guide
level: Build
roles: [marketer]
time: { read: 8, do: 20, wait: 5 }
needs: ["Clean speech audio, or a script to turn into speech", "A front-facing face with a visible mouth"]
last_verified: 2026-11-11
models_checked: ["Sync-3 Lipsync", "Veed Lipsync v2", "Sync Lipsync React-1", "Minimax H3 Max Lip Sync", "Omni Human 1.5", "Kling AI Avatar 2 (Pro)", "Sync-3 Avatar Image to Video", "Sync Lipsync v2", "Veed Fabric Lipsync 1.0", "Kling LipSync"]
related: [narration-and-speech, dub-and-translate-video, presenter-video-with-avatar, generate-video, video-tools]
steps:
  - title: "Prepare the speech"
    text: "Clean speech with no music under it, at a natural pace, made with a narration model and listened to first. Trim any silence before the first word and compare the audio's length with the video or the model's limit."
  - title: "Prepare the face"
    text: "The mouth fully visible, even light on the face, room above the head and below the chin for a portrait, and little camera movement in a source clip. Generate and approve the portrait first, at the framing you want."
  - title: "Open the video panel and pick the model"
    text: "Create › Videos, click the Model card › View all models, search \"lip sync\" or the model's name, read its inputs on the right and click Use this model."
  - title: "Re-sync an existing clip with Sync-3 Lipsync"
    text: "Trim the source to the spoken part with Video Cut, choose Sync-3 Lipsync, load the video and the audio, and pick the sync mode: keep cut_off when both are the same length, remap when the full video and full audio must both stay. Check the price on the button and generate."
  - title: "Animate a portrait"
    text: "Choose Minimax H3 Max Lip Sync, load the approved portrait and 5 to 15 seconds of audio (no text prompt), and pick a lower resolution for a test, higher for the final take. Split longer scripts at sentence breaks, one clip per part."
  - title: "Check the mouth"
    text: "Watch at full speed with sound, then step through the frames on the hard sounds. Accept when the lips close on \"p\", \"b\" and \"m\" and the face stays the same person; reject a mouth that lags or leads, flickering teeth or a blurred mouth area."
agent_steps:
  - title: "Connect the Scenario MCP"
    text: "Connect your agent and sign in once at [mcp.scenario.com](https://mcp.scenario.com). Then tell it which Scenario team and project to use, so every file lands there."
  - title: "Give the agent the files"
    text: "The clean speech (no music under it, no silence before the first word), and either the source clip to re-sync or the approved front-facing portrait. Upload them with upload_asset or give their asset IDs. Use invented faces, or a real person's only with their consent."
  - title: "Brief the agent: re-sync a clip"
    prompt: |-
      Trim my source clip to the spoken part with Video Cut (model_scenario-video-cut). Then re-sync the mouth to my audio with Sync-3 Lipsync (model_sync-lipsync-v3): the trimmed video and the audio, sync mode cut_off if they are the same length, remap if both must stay whole (read model_schema_get for the parameter names). Tell me both lengths before you run. Quote the cost with a dry run first and wait for my go.
  - title: "Or brief the agent: animate a portrait"
    prompt: |-
      Animate my approved portrait with Minimax H3 Max Lip Sync (model_minimax-h3-max-lip-sync): the portrait and the audio, no text prompt, a low resolution for this test (read model_schema_get for the parameter names and resolutions). The audio must be 5 to 15 seconds: if it is longer, split it at sentence breaks and make one clip per part with the same portrait. Quote the cost with a dry run first and wait for my go.
  - title: "Check and redo"
    text: "Accept when the lips close on \"p\", \"b\" and \"m\", the mouth opens on vowels as you hear them, teeth stay stable and the face stays the same person. Reject a mouth that lags or leads, lips open on closing sounds, flickering teeth, a changed face, or a blurred mouth area. Fix the inputs before you try another model."
    prompt: |-
      Wait for the job with jobs_wait and show me the clip. Extract the frames around each "p", "b" and "m" with Video to Image Sequence (model_scenario-video-to-image-seq) and tell me where the lips stay open. Redo only on my go.
  - title: "File and report"
    prompt: |-
      Create a collection for this presenter with the portrait or source clip, the audio and the accepted clips, and report the models, the settings and every asset ID.
updated: 2026-11-11
canonical: https://learn.scenario.com/video/lip-sync-and-talking-avatars
---

Last updated: 2026-11-11

# Lip sync video and make talking avatars

> Sync new speech to an existing clip, or animate a portrait into a presenter from an audio track, and check mouth shapes and identity.

In short:

- Re-sync the mouth of an existing clip to new speech, or animate one portrait into a talking presenter.
- Clean, natural speech and a well-lit, visible mouth decide most of the result: fix inputs before models.
- Find both kinds of model in the model library of Create > Videos; runs are billed by length, so trim the source first.


Lip sync covers two jobs. **Re-sync** takes a clip of someone speaking and rewrites the mouth to new audio: a corrected line, another language, another voice. **Talking avatar** takes one still portrait and an audio track and returns a clip of that face speaking, with expressions, blinks and head movement. This guide prepares the speech and the face, runs each job with its main model, and shows how to check the mouth frame by frame.

:::note As of November 2026
The models, settings and prices in this article are the ones Scenario offered in November 2026. New models and updates arrive often, so check the model library for newer options before you choose.
:::

:::steps Quick version
1. Make clean speech at a natural pace, with no music under it, and listen to it first.
2. Choose the face: a front-facing portrait, or a source clip with the mouth visible and little camera movement.
3. Open **Create** › **Videos**, click the **Model** card, then **View all models**, and pick a lip sync model.
4. To re-sync a clip, use Sync-3 Lipsync with the video, the audio and a sync mode. For a portrait, use Minimax H3 Max Lip Sync with the image and the audio.
5. Check the price on the button, generate, and check the lips on "p", "b" and "m".
:::

## Before you start

| You have | You get | Model to start with |
| --- | --- | --- |
| A clip of someone speaking, plus new audio | The same clip with the mouth rewritten to the new audio | Sync-3 Lipsync |
| One image of a face, plus audio | A talking-head clip: mouth, expressions, head movement, blinks | Minimax H3 Max Lip Sync |

- **Rule of thumb from the Knowledge Center:** if you already have a video of the character, re-sync it, so the original motion and camera stay. If you only have a still, animate the portrait; the clip runs as long as the audio.
- **The audio decides most of the result.** The face decides the rest. Fix the inputs before you try another model.
- **Lip sync is for speech, not singing.** Sync-3 gives generic mouth shapes on music.
- **Runs are billed by length.** Trim the source to the spoken part first, and check the price on the button.
- **Use invented faces**, or a real person's only with their consent.

:::flow From script to talking clip
Speech | narration model, clean, natural pace | audio
Face | approved portrait or source clip | image
Sync | lip sync model | video
Check | closing sounds, identity |
:::

![Two lip sync jobs. Animate a portrait: write the script and split long scripts at sentence breaks; make the speech with a narration model and listen to it first; generate and approve a front-facing portrait with the mouth visible and room above the head; send both to a talking-head model in the video panel, within its audio length limit. Re-sync an existing clip: trim the source clip to the spoken part with Video Cut, add clean new speech with no music under it, and run a lip sync model whose sync mode decides what happens when the lengths differ. Both end with the same check by eye: lips close on p, b and m, and the face stays the same person throughout.](https://ufgc0c9mhl4ueq76.public.blob.vercel-storage.com/academy/d26a7047490e62dd104b1967b115351e734c6f8ae86f63b556d9a283369c1328.svg "Animate a portrait or re-sync a clip, then check the mouth by eye.")

## Step 1: Prepare the speech

- **Clean speech.** No music or effects under the voice. If the track has a bed, separate the voice first, sync, then remix.
- **Natural pace.** Clipped or rushed text-to-speech gives choppy mouth movement. Generate speech with a narration model and listen to it first. See [Narration and speech](https://learn.scenario.com/audio/narration-and-speech.md).
- **More expression.** With text-to-speech, exclamation marks and capital letters in the script give stronger delivery, and the mouth follows it.
- **No silence at the start.** A pause before the first word can push the sync off. Trim it.
- **Know the length.** Compare the audio with the video or the model's limit, and decide what happens when they differ (Step 4).

## Step 2: Prepare the face

- **Mouth fully visible:** front-facing or three-quarter, no hand or microphone over it. Sync-3 detects hands and microphones near the mouth, but a clear mouth still gives the sharpest result.
- **Even light on the face.** Deep shadows around the mouth make teeth and lips smear.
- **Room around the head** for a portrait, above the head and below the chin: the model adds head movement. Head and shoulders at eye level, on a plain background, gives the model the most to work with.
- **Little camera movement** in a source clip. Fast moves and cuts make the sync drift.
- **Generate the portrait first** with an image model, at the framing you want, and approve it like any first frame. The output keeps the image's framing and aspect ratio.

## Step 3: Open the video panel and pick the model

From the top menu, choose **Create**, then **Videos**. Click the **Model** card, then **View all models** to open the library, and search for the model by name, or for "lip sync" to see them all. Click one to read its inputs in the panel on the right, then click **Use this model**. For a walkthrough of the library, see [Generate video](https://learn.scenario.com/video/generate-video.md).

## Step 4: Re-sync an existing clip with Sync-3 Lipsync

Sync-3 Lipsync re-syncs the mouth in any existing video to a new audio track, for dubbing, voice-over and fixing a line. It processes the whole frame (no face crop needed), outputs up to 4K, and can open the lips of a character who wasn't speaking in the source.

:::steps
1. Trim the source video to the spoken section with Video Cut. See [Video tools](https://learn.scenario.com/video/video-tools.md).
2. Choose **Sync-3 Lipsync** in the library.
3. Load the video and the audio.
4. Pick the sync mode. Keep `cut_off` when the two are the same length.
5. Check the price on the button and generate.
:::

The sync mode decides what happens when the video and the audio aren't the same length:

| Sync mode | What happens when lengths differ | Use it for |
| --- | --- | --- |
| `cut_off` (default) | The audio is trimmed to the video's length | A fixed video length, when the end of the audio doesn't matter |
| `loop` | The video repeats from the start to cover the audio | Short ambient clips, looping background characters |
| `bounce` | The video plays forward, then backward | Idle animations, walk cycles |
| `silence` | Silence is added to the audio to cover a longer video | The subject stops speaking partway through |
| `remap` | The video's playback speed changes to fit the audio | Both the full video and the full audio must stay |

Other models that re-sync a clip: **Sync Lipsync v2** (keeps the speaker's style, handles several speakers), **Sync Lipsync React-1** (fast), **Veed Lipsync v2** and **Kling LipSync** (short, quick clips; also takes text-to-speech). The Knowledge Center notes that the Sync models work best on realistic faces.

## Step 5: Animate a portrait

**Minimax H3 Max Lip Sync** takes one portrait and 5 to 15 seconds of audio, and no text prompt: the image and the audio drive everything. It outputs at 480P, 768P, 1080P or 2K, the aspect ratio always follows the image, and it works on photos, illustrated characters and stylized 3D or mascot art, in any spoken language.

:::steps
1. Choose **Minimax H3 Max Lip Sync** in the library.
2. Load the approved portrait and the audio (5 to 15 seconds).
3. Choose the resolution: lower for a test, higher for the final take.
4. Check the price on the button and generate.
:::

Other portrait models:

| Model | Audio length | Notes |
| --- | --- | --- |
| Minimax H3 Max Lip Sync | 5 to 15 s per clip | No text prompt; 480P to 2K; aspect ratio follows the image |
| Omni Human 1.5 | Under 30 s | Natural gestures and emotion; photoreal and stylized faces |
| Kling AI Avatar 2 (Pro) | See model page | Image or video plus audio; real or stylized portraits |
| Sync-3 Avatar Image to Video | Output length equals the audio | No prompt or settings; photos, illustrations, anime, 3D renders |
| Veed Fabric Lipsync 1.0 | See model page | Quick talking heads that keep the image's style |

**Split long scripts** at sentence breaks when they're longer than the model's limit. Make one clip per part with the same portrait, then join them. See [Presenter video with avatar](https://learn.scenario.com/playbooks/presenter-video-with-avatar.md).

## Step 6: Check the mouth

Watch the clip at full speed with the sound on, then step through the frames on the hard sounds. Lips must close on "p", "b" and "m". Watch it twice more: once for the face, once for the teeth.

:::check
**Accept when:** the lips close on "p", "b" and "m", the mouth opens on vowels at the moment you hear them, teeth and tongue look stable, the face stays the same person throughout, and the head movement looks natural for the voice.
**Reject when:** the mouth lags or leads the audio, lips stay open on closing sounds, teeth flicker or merge, the face changes shape or age, or the mouth area looks blurred against the rest of the face.
:::

## Make it repeatable

- **One portrait and one voice per presenter:** keep the approved portrait (or source clip) and the voice setting in a collection.
- **A workflow per series:** it takes a script, makes the speech and passes both to the lip sync model, so each new video is one run with the same face and voice.

## Common problems

| What you see | What to do |
| --- | --- |
| The sync is off by a fraction of a second | Check for silence at the start of the audio. Trim it and run again |
| The mouth smears or flickers | Improve the source: a sharper face, even light, less camera motion. Then try another model from the same table |
| The clip stops before the audio ends | The mode is `cut_off`, or the model's limit was reached. Choose `loop` or `remap`, or split the audio |
| The mouth moves vaguely on a song | Lip sync models are made for speech. Use a spoken track |
| Stylized faces lose their style | Some models are tuned for photoreal faces. Try Omni Human 1.5, Minimax H3 Max Lip Sync or Sync-3 Avatar Image to Video |
| The wrong face speaks in a group shot | Use a model with active speaker detection, such as Sync Lipsync v2, or crop to one face first |
| The job is refused by moderation | Use invented faces, or a real person's only with their consent |

## What's next

To say the same thing in other languages, go to [Dub and translate video](https://learn.scenario.com/video/dub-and-translate-video.md). To make the speech, see [Narration and speech](https://learn.scenario.com/audio/narration-and-speech.md).

Source: https://learn.scenario.com/video/lip-sync-and-talking-avatars