HiNow LogoHiNow

MultiTalk

Model Information

Display Name: MultiTalk

API Model ID: infinitetalk/multitalk

Category: Audio To Video

Description: MultiTalk is an advanced audio-driven talking and singing video generation model that creates realistic videos from a single portrait image and an audio clip. Built on Wav2Vec Audio Encoder combined with Wan2.1 Video Diffusion and Uni3C ControlNet, it delivers precise lip synchronization, natural body movements, and dynamic camera control. ## Key Features - **Perfect lip synchronization** — precise mouth movements matching every syllable of speech or singing - **Full-body animation** — natural gestures and body language, not just facial movements - **Dynamic camera movements** — powered by Uni3C ControlNet for cinematic shots - **Multi-person support** — handles conversations with multiple speakers - **Long-form content** — supports videos up to 10 minutes from a single generation - **Singing mode** — works with both speech and music/singing audio ## Best Use Cases - AI avatars and virtual presenters for corporate training and marketing - Music video creation — animate characters singing along to tracks - Dubbing and localization — reanimate speakers with translated audio - Podcast and audiobook visualization with talking characters - Social media content with realistic talking-head videos - Multi-person dialogue scenes from audio conversations - Educational content with animated instructors ## Parameters | Parameter | Type | Required | Default | Description | |-----------|------|----------|---------|-------------| | image | URL | Yes | — | Portrait reference image (clear, front-facing, good lighting) | | audio | URL | Yes | — | Audio file URL (up to 10 minutes) | | prompt | string | No | — | Scene control, pose, and behavior guidance | | seed | integer | No | random | Seed for reproducibility (-1 for random) | ## Tips for Best Results - Use clear, front-facing portrait photos with good lighting - Ensure faces are clearly visible without obstructions - Use high-quality audio with minimal background noise - Neutral or slightly open mouth expression in source image works best - For multi-speaker content, provide distinct audio tracks per speaker ## Pricing Cost is calculated per second of audio duration at /bin/bash.042/s. A 10-second clip costs approximately /bin/bash.42, a 1-minute video costs .52.

How to Use This Model

To use MultiTalk via the HInow.ai API, use the model ID: infinitetalk/multitalk

API Request Example


POST https://api.hinow.ai/v1/chat/completions
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

{
  "model": "infinitetalk/multitalk",
  "messages": [
    {"role": "user", "content": "Your input here"}
  ]
}

Pricing

  • total: $0.045

Available Parameters

  • audio: Single audio URL (mp3/wav) — or use left_audio/right_audio for duo mode.
  • resolution: Output resolution. 720p costs 2x per second. (Options: 480p, 720p)
  • left_audio: Audio URL for the left speaker (duo mode).
  • right_audio: Audio URL for the right speaker (duo mode).
  • duo_order: Speaking order for duo mode.
  • seed: Random seed for reproducibility. -1 (or omit) for random.

Quick Reference

To use this model, set: "model": "infinitetalk/multitalk"

Featured: No

Documentation: https://hinow.ai/models/infinitetalk/multitalk

API Endpoint: https://api.hinow.ai/v1

Back to Models

MultiTalk

infinitetalk/multitalk

$0.045
per second

About

MultiTalk is an advanced audio-driven talking and singing video generation model that creates realistic videos from a single portrait image and an audio clip. Built on Wav2Vec Audio Encoder combined with Wan2.1 Video Diffusion and Uni3C ControlNet, it delivers precise lip synchronization, natural body movements, and dynamic camera control.

Key Features

  • Perfect lip synchronization — precise mouth movements matching every syllable of speech or singing
  • Full-body animation — natural gestures and body language, not just facial movements
  • Dynamic camera movements — powered by Uni3C ControlNet for cinematic shots
  • Multi-person support — handles conversations with multiple speakers
  • Long-form content — supports videos up to 10 minutes from a single generation
  • Singing mode — works with both speech and music/singing audio

Best Use Cases

  • AI avatars and virtual presenters for corporate training and marketing
  • Music video creation — animate characters singing along to tracks
  • Dubbing and localization — reanimate speakers with translated audio
  • Podcast and audiobook visualization with talking characters
  • Social media content with realistic talking-head videos
  • Multi-person dialogue scenes from audio conversations
  • Educational content with animated instructors

Parameters

ParameterTypeRequiredDefaultDescription
imageURLYes—Portrait reference image (clear, front-facing, good lighting)
audioURLYes—Audio file URL (up to 10 minutes)
promptstringNo—Scene control, pose, and behavior guidance
seedintegerNorandomSeed for reproducibility (-1 for random)

Tips for Best Results

  • Use clear, front-facing portrait photos with good lighting
  • Ensure faces are clearly visible without obstructions
  • Use high-quality audio with minimal background noise
  • Neutral or slightly open mouth expression in source image works best
  • For multi-speaker content, provide distinct audio tracks per speaker

Pricing

Cost is calculated per second of audio duration at /bin/bash.042/s. A 10-second clip costs approximately /bin/bash.42, a 1-minute video costs .52.

Capabilities

Audio To Video

Parameters

audio

Single audio URL (mp3/wav) — or use left_audio/right_audio for duo mode.

resolution

Output resolution. 720p costs 2x per second.

480p720p
left_audio

Audio URL for the left speaker (duo mode).

right_audio

Audio URL for the right speaker (duo mode).

duo_order

Speaking order for duo mode.

seed

Random seed for reproducibility. -1 (or omit) for random.

Code Examples

curl -X POST https://api.hinow.ai/v1/videos \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $HINOW_API_KEY" \
  -d '{
    "model": "infinitetalk/multitalk",
    "prompt": "A cat playing with a ball",
    "parameters": {
      "audio": "",
      "resolution": "480p",
      "left_audio": "",
      "right_audio": "",
      "duo_order": "",
      "seed": ""
    }
  }'