InfiniteTalk · Image to Video

ABOUT THE WORKFLOW

Animate a Face to Any Audio Length
Upload a portrait image and an audio file. The workflow separates vocals from background noise, extracts speech features, and generates a talking avatar video where the character's mouth syncs to your audio. InfiniteTalk processes the audio in overlapping windows and loops until the full recording is covered, so there is no duration cap. A 10-second clip and a 5-minute monologue use the same workflow. The output is an MP4 with your original audio embedded.

Model

• Wan 2.1 I2V 14B by Alibaba with the InfiniteTalk Multi adapter. Generates audio-driven talking head video in windowed chunks that tile across the full audio length, with built-in vocal separation and Wav2Vec2 feature extraction for accurate lip-sync across languages.

HOW IT WORKS

Step 1. Upload your image
A portrait or headshot of the person you want to animate. Front-facing, well-lit, with a visible mouth and jaw.
Works great with: portraits · headshots · AI-generated faces · character renders

Step 2. Upload your audio
A recording of speech, singing, or voice acting at any length. The workflow separates vocals from background noise automatically before processing.
Works great with: voiceovers · podcast clips · dialogue recordings · singing · narration

Step 3. Write a short prompt
Describe the action and mood. "A man is talking" or "A woman singing with gentle expression" tells the model what motion style to generate around the lip-sync.

Step 4. Hit run and download
InfiniteTalk processes the audio in windowed segments, generates lip-synced video for each, and combines everything into a single MP4 at 25fps with your audio.
Ready for: Premiere · DaVinci Resolve · After Effects · TikTok · YouTube · podcasts

First time? Upload a clear front-facing portrait and an audio clip. Leave every setting as-is. The defaults are the right starting point for almost everyone.

RECOMMENDED SETTINGS

Quick-start guide. Find the goal that matches yours and copy the settings.

• Standard talking avatar video — Default settings (854x480, 81-frame windows, 25fps, 6 steps). Upload your image and audio and run.

• Longer audio (minutes, not seconds) — No changes needed. InfiniteTalk loops through the audio in windows until the full file is covered. Generation time scales linearly with audio length.

• Singing instead of speech — Change the prompt to describe singing. "A woman singing softly with subtle head movement" steers the motion toward a musical performance.

• Stronger lip-sync effect — Increase the audio_scale value in the Wav2Vec node. Higher values make the mouth track the audio more aggressively. Lower values produce subtler movement.

• Lip-sync looks off or jittery — Start with cleaner audio. The workflow includes vocal separation via MelBandRoFormer, but heavily mixed or reverb-heavy tracks still introduce noise into the feature extraction. A clean voice recording produces the best sync.

• Face looks distorted — Use a higher-quality, front-facing portrait with even lighting. The mouth and jaw must be fully visible with no occlusion from hands, hair, or accessories.

Prompt: Keep it short and focused on the performance type. "A man talking calmly with slight head nods" is specific. "A person in a room" adds nothing useful.

LEARN

📹 Videos

• Intro to Floyo

• ComfyUI 101 Free Course ft. Sebastian Kamph

• Floyo 101 for Team Collaboration

✨ Quick links

• Help docs

• LoRA Training guide

• Discover more workflows

USE CASES

🎙️ Podcast and Narration Visuals
Turn a full podcast episode or narration recording into a talking avatar video, giving audio-only content a visual presence for YouTube and social distribution.

🎤 Audio-Driven Talking Head Videos
Upload a voiceover of any length and a portrait to produce a talking head clip where the character's mouth syncs to the full recording.

🎵 Singing and Music Performance Clips
Upload a vocal track and a character image to generate a singing performance video with matched lip movement for the entire song.

👤 Virtual Spokesperson for Course Content
Generate a consistent talking character from a single portrait across full-length explainer videos, tutorials, or training modules without recording on camera.

📱 Long-Form Social Content
Produce talking avatar videos longer than what other lip-sync tools allow, covering full product walkthroughs, story times, or monologues for YouTube and TikTok.

WHAT WORKS BEST / WHAT TO AVOID

✅ Works great

• Front-facing portraits with visible mouth, jaw, and chin

• Clean audio with isolated speech or vocals

• Audio of any length, from seconds to minutes

• Short prompts describing the performance mood

⚠️ May produce softer results

• Profile or three-quarter angle portraits where the mouth is partially hidden

• Audio with heavy background music, reverb, or overlapping voices

• Portraits with hands, objects, or hair covering the lower face

• Very low-resolution or blurry source images

FAQ

What is InfiniteTalk and how is it different from MultiTalk?
InfiniteTalk is an extension of the MultiTalk lip-sync adapter for Wan 2.1. Standard MultiTalk generates a fixed number of frames in a single pass, which caps the video duration. InfiniteTalk processes the audio in overlapping windows (81 frames each at 25fps), generates video for each window, re-encodes the frames, and loops to the next segment. This removes the duration cap entirely. A 10-second clip and a 10-minute monologue use the same pipeline.

Is there a maximum audio length for InfiniteTalk?
No hard limit. The workflow loops through the audio in windowed chunks until the full file is covered. Generation time scales linearly with audio length. Longer audio takes proportionally longer to render, but the pipeline handles it automatically.

Does InfiniteTalk lip-sync work in languages other than English?
Yes. The Wav2Vec2 encoder reads acoustic features from the audio signal, not language-specific text. Lip-sync works across languages including English, Mandarin, Japanese, Korean, Spanish, and others. The audio drives the mouth movement regardless of language.

Does the workflow separate vocals from background music automatically?
Yes. The workflow includes a MelBandRoFormer vocal separation step that isolates the voice from background noise and music before extracting speech features. Starting with clean audio still produces better results, but mixed tracks are handled.

Can I use my own voice recording as the audio input?
Yes. Record your voice, upload the file, and the model syncs the character's mouth to your speech. Any standard audio format works. The workflow handles the rest.

Is InfiniteTalk licensed for commercial use?
InfiniteTalk is built on Alibaba's open-source Wan 2.1 model. The InfiniteTalk adapter, LoRAs, and supporting models each have their own license terms. Check each component's license on its model page for commercial use in your specific project.

How to run InfiniteTalk lip-sync video generation online?
You can run InfiniteTalk lip-sync video generation online through Floyo. No installation, no setup, no local GPU needed. Open the workflow in your browser, upload your image and audio, and hit run. Free to try.

WHY FLOYO?

Floyo is the only platform with team collaboration for ComfyUI in the browser. You run workflows with no install. You share run history, assets, and models across your team. You pay only when you generate. Floyo supports open-source and closed-source models.

A designer runs an edit and likes the result. A teammate opens that exact run from shared history and keeps going. No file handoffs. No version confusion.

For studios and enterprise teams, Floyo adds private workspaces, pooled resources, and a team usage dashboard. Other ComfyUI cloud tools run for one person at a time. Floyo runs for the whole team, with transparent per-generation costs.

Ready to try it?
Upload a portrait and an audio file of any length, and hit run.

→ Launch Workflow, Free

Questions? Watch the free course or check the FAQ above.