Skip to content
Get in touch

Lip Sync

Start with a hosted service that can animate an original character image, presenter, or authorized existing video with speech. Explore manual timing or a rigged character when you need more control.

Go to My experiment

Explore

What are the ways to do this?

Worked example brief: an illustrated host says one welcome line

Use an original, non-real character and an authorized voice for a 10–12 second repair-café welcome. Keep the exact line, framing, captions, and disclosure fixed.

Notice: Which path gives enough timing and correction control while protecting identity and permission?

Try it with your own tools

We have provided a brief for you to try; we have not produced a finished example for this task.

What this shows: How to bound a lip-sync comparison around one original character, one authorized line, and explicit disclosure.

What it cannot show: How a service performs, how viewers interpret the clip, or whether a real person has consented to another project.

  • This is a worked brief, not a published clip or hands-on service trial.
  • No sync quality, identity stability, or export result is claimed.
  • A real trial must retain character and voice permissions, inputs, settings, corrections, outputs, captions, and disclosure.

Approach map

Open any approach to inspect its starting material, human work, outputs, and sequence.

Start with a hosted lip-sync servicePair an original character image or presenter, or an authorized existing video where supported, with speech to create a short first pass.
Starts with
An accepted image, presenter, or existing-video input · Authorized recorded or stock synthetic speech · Current upload, retention, correction, and export terms
Main actions
Prepare a short bounded input · Generate one pass · Inspect identity and timing · Revise one visible mismatch
Outputs
Rendered clip and captions · Input type, settings, correction, and disclosure record
Human work
Confirm permission and terms · Judge every frame · Correct, edit, or reject
Useful when
A first lip-sync attempt should be quick to inspect · The service exposes enough correction or export control
  1. Choose the accepted inputCheck whether the service starts from a character image, presenter, or existing video; those paths are not interchangeable.
  2. Use one short authorized lineKeep the first attempt brief and avoid sensitive likeness or voice material.
  3. Inspect and revise one phraseReview closures, pauses, teeth, face edges, and background drift, then make one targeted correction.
  4. Finish with captions and disclosureKeep the input type and production method clear.
Take more control with manual mouth timingPlace a small set of mouth drawings, poses, or keyframes against the waveform when direct timing control matters.
Starts with
Authorized audio · Original or permitted character art · A timeline with waveform access
Main actions
Mark phrases and closures · Place mouth shapes · Preview at normal speed · Adjust the most visible mismatches
Outputs
Editable animation timeline · Rendered clip and caption file
Human work
Interpret the speech · Draw or choose mouth shapes · Protect character consistency
Useful when
A short line needs frame-level control · An editable source file matters more than a quick first pass
  1. Confirm the inputsRecord who made and authorized the character and voice.
  2. Mark the waveformStart with pauses, strong closures, and broad vowel openings.
  3. Place a small mouth setAim for readable rhythm before fine detail.
  4. Review and captionCheck timing, identity stability, disclosure, and captions.
Use a rigged character for reusable controlMap speech analysis, live performance, or phoneme cues onto a character rig when repeated lines justify more setup.
Starts with
A character rig you may use · Authorized speech or performance · Known mouth-shape controls
Main actions
Map or capture timing · Tune poses and transitions · Edit the performance · Render and caption
Outputs
Editable performance data · Rendered clip and source record
Human work
Prepare the rig · Direct performance · Fix timing and expression · Maintain disclosure
Useful when
A recurring character needs reusable control · Future line changes are likely
  1. Prepare the character rigTest the mouth set and neutral pose before speech.
  2. Choose the driverKeep the source of motion clear.
    • Speech analysisMap cues to poses: Correct transitions after the automatic pass.
    • Live performanceCapture with consent: Record performer permission and keep editable data.
  3. Edit, caption, and discloseKeep the source performance and final claims accurate.

Start with your intention

Use My experiment to note your audience, material, limits, equipment, and sharing restrictions. Every field is optional. “Try with our example” fills only an empty goal and constraints field.

Prompt · version 1

Investigate ways to create lip sync

When an authorized character needs to appear to speak and you need to understand the workflow before uploading face or voice material.

Map manual, rigged, and service-driven lip-sync workflows with consent and correction control visible.

Example: An original illustrated host says one 10–12 second repair-café welcome line with captions and disclosure.

Show the full prompt

The published prompt stays unchanged.

Source checked ·

Tools to investigate

Short novice lip-sync work is feasible when the scene is simple, the face is clear, the audio is clean, and the person or character is authorized. The four tools represent preset or uploaded characters, stock or custom avatars, relip-syncing existing footage, and controllable 2D puppets. They are not interchangeable identity workflows. This is a source-checked starting shortlist, not a ranking or a hands-on comparison.

This shortlist reflects the date above. Confirm current features, terms, pricing, privacy, exports, and accessibility before choosing.

A high-level workflow

  1. Choose the identity model first: a service-provided stock character, an owned fictional character, a consented custom avatar, an existing consented talking clip, or a reusable 2D puppet.
  2. Use a 5–10 second script, one visible speaker, clean speech, a front-facing well-lit source, and no mouth obstruction for the first trial.
  3. Review every frame around the mouth, teeth, jaw, blinks, and cuts; confirm voice and likeness permission, disclose synthetic performance where appropriate, and keep the original assets and consent record.

Runway Character Script to Video

Purpose and fit
A short talking fictional or preset character generated from one centered image plus typed dialogue or owned audio.
How to explore it
Choose a preset character or upload an owned character image, provide a short script with a preset voice or clean audio, and review the generated performance before any edit or publication.
Watch for
The old standalone Runway Lip Sync tool is deprecated. The current app uses credits; uploaded faces and cloned voices require rights and consent, and the result still needs frame-level review.

Sources checked

HeyGen Studio

Purpose and fit
Presenter-style scenes using a public stock avatar, or a recurring authorized custom Digital Twin when the added identity setup is justified.
How to explore it
For a low-risk first trial, choose a public avatar, add a short script or owned audio, preview the voice, and render one scene. Create a Digital Twin only after the depicted person completes the required consent process.
Watch for
A public avatar and a custom Digital Twin have different identity, privacy, and setup implications. HeyGen requires a consent video for video-based Digital Twins; subscription, export, disclosure, and reuse terms need checking.

Sources checked

Sync Labs Studio

Purpose and fit
Replacing the speech in an existing consented talking-head clip while retaining the underlying performance and comparing dedicated lip-sync models in a browser.
How to explore it
Use Studio's no-code path with one short front-facing talking clip and a clean, similarly timed audio file; compare a general model before considering higher-cost models for difficult angles or obstructions.
Watch for
Some models need natural speaking motion in the source and can fail on static frames, profiles, obstructions, multiple faces, or duration mismatches. Free or lower plans may watermark or limit duration; uploaded likeness and audio require permission.

Sources checked

Adobe Character Animator

Purpose and fit
A stylized 2D puppet that should remain editable and reusable, with lip sync driven by microphone input or prerecorded scene audio.
How to explore it
Start from a properly rigged puppet with mouth shapes, add or record clear audio, compute the lip-sync take, then correct timing or mouth shapes in the scene before export.
Watch for
This is puppet animation, not photorealistic relip-sync. It has more asset setup and desktop-tool learning than a stock-avatar service, and poor audio or missing visemes reduce accuracy.

Sources checked

Decide

What matters for your project?

Use a few criteria and a separate must-have list. An unknown must-have remains unresolved.

Must-have

Consent and identity

Character, face, voice, and performance inputs are authorized.

Acceptable: The record names who authorized each input and forbids an implied real endorsement.

Must-have

Speech timing

Mouth movement supports closures, openings, pauses, and phrase rhythm.

Acceptable: The welcome line reads coherently at normal speed without a distracting mismatch.

Must-have

Character stability

Face, mouth, teeth, framing, and background remain intentional.

Acceptable: No identity or visual drift distracts from the line.

Preference

Correction control

A maker can repair one phrase or pose without rebuilding the whole clip.

Acceptable: Source files or controls support a targeted correction and record the effort.

Must-have

Disclosure, captions, and export

The clip is captioned, honestly described, and usable in the intended destination.

Acceptable: Captions, method disclosure, editable source where available, and a common export are retained.

AI reported

What an AI response says. Follow the sources before relying on it.

Source checked

What an opened source states. This does not prove product performance.

We tested

What RS observed in a dated, bounded trial.

Still uncertain

What is unknown, unverified, or outside the trial.

Prompt · version 1.0

Shape lip-sync criteria one question at a time

Before choosing an automatic service, rig, or manual animation path.

Turn a speaking-character idea into consent-aware, testable criteria.

Example: The clip must use an original character and authorized voice, stay visually stable, include captions, and disclose the method.

Show the full prompt

The published prompt stays unchanged.

Prompt · version 1.0

Compare two lip-sync paths without hiding identity risk

After creating two authorized passes or revising one phrase once.

Compare sync and correction workflows while keeping permissions, captions, and disclosure in view.

Example: Compare manual mouth timing with a service-driven pass for the same illustrated host and welcome line.

Show the full prompt

The published prompt stays unchanged.

Experiment

Make one short hosted-service pass and revise one phrase

Start with a hosted service that accepts a permitted character image or presenter, or an existing video where supported, together with authorized speech. Make one bounded pass, then improve one visible mismatch.

Ready-to-use brief

Create a 10–12 second clip of an original illustrated, non-real host saying: “Welcome to the repair café. Bring one object, and we’ll ask what it needs.” Keep the synthetic or recorded voice authorized, disclose the method, and provide the line as captions.

Input

Use one original character image, presenter design, or existing clip you may animate and one voice recording or stock synthetic voice you may use. Check which input the chosen service accepts. Do not use a real person’s face or cloned voice for this trial.

Limit

Aim for 45–60 minutes for the first pass and one targeted revision, with no purchase. The published page provides a worked brief only; it does not claim that a lip-sync output was produced or reviewed.

Path 1

Image or presenter path: pair an original character image with authorized speech in a hosted service.

Path 2

Existing-video path, if supported: pair an authorized clip with speech and record how that different starting input affects the workflow.

Only one candidate? Create one hosted-service pass, correct one phrase with the available controls or a small follow-up edit, and compare the before and after timing. A second service path is optional; disclose its different input rather than treating the outputs as a direct ranking.

Define success before you begin

  1. Confirm the service accepts the intended image, presenter, or existing-video input, and check permission plus current upload, retention, and export terms.
  2. Preserve the exact audio, character source, captions, and disclosure.
  3. Create one short hosted-service pass and save its settings and output.
  4. Choose one visible timing mismatch and make one targeted revision.
  5. Review at normal speed and frame by frame around the revised phrase.
  6. Record the input type, correction, effort, cost, exports, and unknowns; document an optional second path separately.

Observation questions

  • Do mouth closures, openings, and pauses support the spoken line?
  • Does the character remain visually stable and authorized?
  • Can one mismatched phrase be corrected without rebuilding the clip?
  • Are captions and a clear synthetic/animated disclosure present?
  • What files, settings, consent, retention, and export limits need recording?

What would you do next? Continue, change an approach, check a claim, try a simpler method, or stop.

Recorded on 2026-09-18

We ran the landscape prompt

Current official-source feasibility review covering Runway, HeyGen, Sync Labs, and Adobe Character Animator with identity, consent, and workflow distinctions.

Read the prompt, response, sources, evidence labels, limits, and next question.