---
title: "Discussion on \"Conditioning a video model with audio: three reference clips, and where lip sync actually stops\" | Hashnode"
description: "Discussion on \"Conditioning a video model with audio: three reference clips, and where lip sync actually stops\". H3 accepts up to three audio references. What that buys you on each lane, and the one thing it does not promise"
url: "https://hashnode.com/posts/h3-reference-audio-conditioning/6ae6f6c173e9f4171f4e133a"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [Conditioning a video model with audio: three reference clips, and where lip sync actually stops](https://callirra.hashnode.dev/h3-reference-audio-conditioning?utm_source=hashnode&utm_medium=feed)

H3 accepts up to three audio references. What that buys you on each lane, and the one thing it does not promise

Nov 1 · 5 min read · callirra.hashnode.dev  
Read the article

-   [#minimax](https://hashnode.com/tag/minimax)
-   [#image-to-video](https://hashnode.com/tag/image-to-video)
-   [#audio](https://hashnode.com/tag/audio)
-   [#video-generation](https://hashnode.com/tag/video-generation)

## Written by

[Callirra](https://hashnode.com/@callirra)

@callirra

  
[View profile](https://hashnode.com/@callirra)

One workspace for AI image, video and audio

Callirra is one workspace for AI image, video and audio: 50 media models, 11 multimodal chat models, prompt polish, a private asset library, saved Constellation workflows and AI Filmmaker. The team writes here about real credit costs, model comparisons and release notes. Product and pricing: callirra.com

## Responses 1

Markdown supported · ⌘↵ to post

[Camille Dubois](https://hashnode.com/@camille-dubois)

[Nov 1](https://hashnode.com/posts/h3-reference-audio-conditioning/6ae6f6c173e9f4171f4e133a/comment/6ae6f8e6849868cc5538463f)

Three reference clips is more than I expected. On mobile we mostly care about whether the audio survives the encode - does the model re-render the audio track, or just condition on it and pass it through?
