MiniMax H3 Test: Achieves Accurate Voice Cloning via Audio Reference Workflow

Perfect-Campaign9551 · reddit · 2026-08-05

A Reddit user shared their experience testing the default ref2vid workflow in ComfyUI using the MiniMax H3 model.

Along with an image reference, the user added an interview audio recording as an audio reference, prompting the AI to mimic the voice's tone. The test showed that the model successfully achieved voice cloning, with the user praising its powerful generation capabilities.

Related event: MiniMax H3 Video Model Integrated into ComfyUI for Voice Cloning(2 posts)→

Original post →

More from Multimodal

Multimodal channel →