sync. labs ChatGPT plugin

Use sync. labs in ChatGPT to lip-sync a video, make a photo speak, translate and dub a video, or change spoken words while keeping the original voice. Work through chat or open the embedded Studio to choose media, edit inputs, review costs, and play or download results.

Connect the sync. labs ChatGPT plugin

To change accounts or workspaces, use Change account or workspace in the account menu for instructions, then reconnect through ChatGPT’s Sync connection settings using the intended account and workspace. Signing in to Sync in another tab does not change the existing ChatGPT connection. Reopen Sync after changing identity.

Choose a workflow

Create lip sync from an image or video

1. Choose lipsync video or image to video, then upload a visual or select one from your assets.
2. Add an audio file, record audio, extract audio from a video, or enter a script and choose an available voice. Review generated speech before using it for the video.
3. Choose a model and the settings available for that model. New lipsync and image-to-video requests default to sync-3; existing drafts retain their selection.
4. Review the estimate, then choose Sync to submit the video.

One generation uses one image or video and one audio performance or script. To replace the full performance with a new script, use this workflow and select a voice; preserving the original voice is a separate Edit Dialogue workflow with targeted-edit limits.

For unequal video and audio lengths, review the available sync modes. cut_off ends at the shorter input; looping or bouncing a shorter video can cover longer audio when offered. Image-to-video follows the audio duration. Recheck the estimate after changing duration handling.

Translate and dub a video

In the embedded Studio, translation and final lip-sync are separate actions:

1. Choose translate & dub and select a video with speech. Wait for its audio to be extracted, or select the source audio you want to translate.
2. Use Choose language in the audio controls to select a target language. Review the translation estimate, then choose Translate.
3. Wait for the translated audio to be saved and listen to it. Translation alone does not generate the final video.
4. Review the lip-sync model, settings, and video estimate, then choose Sync to create the video from the selected visual and translated audio.

You do not need to supply a translated script. If your free translation preview has a duration limit, Studio can offer to translate only the first part of the audio. Review and explicitly confirm the shortened preview, or upgrade for longer input. The unselected remainder is not translated, and the final video’s limits and estimate still apply.

Through chat, you can instead ask: “Translate this video into Spanish and lip-sync the result. Show me the combined estimate before starting.” Chat can use the combined Translate & Dub operation, which tracks dubbing and final rendering under one generation ID. Confirm the target language and source language if needed, then keep checking that same generation. For a supported social link, import the video first and wait for the saved asset. Do not run the combined operation again on audio already translated in Studio.

To make another language version, reuse the original video or source audio and choose the new target language. Review the new translation and video costs before starting.

Edit dialogue while keeping the original voice

1. Choose edit dialogue and select a single-speaker video with clear speech.
2. Wait for transcription, then change or remove words in the timed transcript. Use undo, redo, or restore to revise your edits.
3. Request an audio preview and listen to it. Changes to the transcript require a matching new preview.
4. Review the final video estimate and submit when the preview is ready and eligible for video generation.

Edit Dialogue supports targeted changes to existing speech. It does not replace every line with an arbitrary new script, provide general multi-speaker editing, or expose unrestricted segment/timeline editing. If the preview cannot generate a video in ChatGPT, use Continue in Sync. Preview generation and final video generation can have separate charges.

If a dialogue preview completes only partially, listen to it and review the notice before generating: some requested edits or processing may not have been applied. Resolve or revise those edits if you need all changes included.

Upload media the right way

Use the embedded Studio’s upload action for video, image, or audio files. You can also select saved assets, import a publicly accessible media URL, or record video/audio when ChatGPT and your browser allow camera or microphone access. Review recordings before saving them. Check the upload indicator and wait for the asset to finish saving before generating.

Chat attachments are usable only when ChatGPT can pass the file to Sync. If an attachment is inaccessible, upload it inside Sync instead; a video does not have to be attached to the conversation.

Through chat, supported YouTube, TikTok, Instagram, and Facebook video links can be imported into your project when social imports are enabled for your account. Platform availability and import limits apply. If a link cannot be imported, upload a file instead. A webpage URL is not necessarily a direct media URL accepted by the editor’s URL field.

Work with chat and the embedded Studio

Chat can open an existing project, select saved media, change editable model/language/script settings, refresh assets, and display an existing generation in the open Studio when those controls are available. For example: “Use the audio I just uploaded in this project’s lipsync draft, but don’t generate yet.”

Wait for Studio to show the requested change before submitting. A change can remain pending if the editor is closed or busy, or if another edit needs your attention. If chat and Studio both changed the draft, choose Keep Studio changes or Use chat changes in the editor.

Opening a result or updating a draft does not generate another video. Switching projects, workflows, or results can replace the active editor view; check the destination and current draft first. Resolve pending uploads or uncertain submissions before switching.

Projects, drafts, and history

Create a project or search your existing projects in Sync. Project cards show previews and generation counts; open a project to select assets and inspect its generation history. Rename the project from its title when that control is available.

For a chat-created lip-sync, specify the existing project or ask for a new named project before confirming. Without a destination, the tool can use its integration default project, such as ChatGPT generations. Check the returned project when looking for the result in Studio.

Saved generations belong to the Sync project. Reopening a project or selecting a history item checks access and loads the existing job; it does not submit it again. Processing jobs keep their status, and completed jobs open their result.

Editable and archived drafts can resume in the same browser/profile when local storage is available. Draft recovery is scoped to your account and workspace; it does not synchronize across devices or browser profiles and can be lost if you clear browser data. Project assets and accepted generations remain separate from local drafts.

Check generation progress and download the result

Follow progress in Studio or ask: “Check the status of this Sync generation.” In another conversation, include the project or generation ID. Wait on the existing generation rather than asking for a replacement while it is still processing.

When complete, play the video and use the Source/Synced comparison where available. Download the result, or use the link to open the video if ChatGPT cannot download it directly. If playback fails, retry playback or open the project in Sync without generating again.

Use Edit inputs or an available continuation action to prepare another version. Review its inputs and a fresh estimate before submitting. Some continuations open full Studio. Selecting an existing result or preparing a new draft alone does not start paid work.

Review costs and account limits

The account menu shows your plan and available free allowance, credits, or metered usage. Use Refresh account to reload it. Workspace owners and admins can open billing management in Sync when permitted.

• Review the estimate before each paid action. Speech synthesis, translated-audio previews, dialogue previews, and final videos can be separate operations.
• Translation quotes explain whether translation is included, billed by Sync, or billed separately by your provider. A lip-sync-only quote is not a full translation quote.
• Script-based lip-sync estimates use the replacement speech’s estimated length, not automatically the entire source video’s length. They remain approximate until audio is generated or measured; speech synthesis is disclosed separately.
• Changing the script, audio, language, model, or sync settings can change the estimate. Review the refreshed quote before submitting.
• Missing duration, unavailable pricing, exhausted allowance, insufficient balance, or plan limits can block Sync. Choose eligible media/settings, refresh the failed check, or manage the plan in Sync. Do not assume previews or clones are free.

Write prompts that describe the result

Ask for available voices before choosing one. Where reusable voice cloning is available through chat, explicitly request it, provide a sample you own or have permission to clone, and confirm the sample, name, and applicable cost/plan limits. Reuse the saved voice for later script-based videos. Cloning is separate from generating speech or a video; the embedded voice picker does not guarantee a clone-creation or deletion control.

Understand the current limits

Your Sync workspace’s media, duration, model, language, credit, and concurrency limits apply in ChatGPT. Available controls reflect the connection’s capabilities; a tool deployed on Sync’s server may still need to become available in ChatGPT.

The plugin is for supported video workflows. It does not provide subtitle-only output, general text translation, a standalone MP3 voiceover workflow, Edit Appearance, or a full timeline editor for trimming and mixing music. Use Studio for workflows that require the full editor. A long input is not automatically shortened to fit a free allowance.

Troubleshoot the ChatGPT plugin

Build custom AI workflows with the MCP server

For another AI client or a custom integration, use the sync. labs MCP server. The official ChatGPT plugin connects through OAuth; you do not need to configure a local MCP server to use it.