FishAudioInstantVoiceClone - ComfyUI Built-in Node Documentation - ComfyUI

This node creates a private cloned voice from your audio recordings using the Fish Audio API. You provide one or more audio samples, and the node builds a custom voice that can be immediately used for text-to-speech. It accepts 1 to 20 recordings, with a recommended length of 10 to 30 seconds each and a total limit of 270 seconds.

Inputs

Parameter	Description	Data Type	Required	Range
`files`	Audio recordings for voice cloning. This is a growable input: connect one or more audio items (for example `audio_1`, `audio_2`, …) to provide the voice samples.	AUDIO	Yes	1 to 20 recordings
`enhance_audio_quality`	Enhance reference audio quality before training (default: True).	BOOLEAN	Yes	True False

Note: The total duration of all reference audio combined must be under 270 seconds. If the combined duration reaches or exceeds 270 seconds, the node returns an error.

Outputs

Output Name	Description	Data Type
`voice`	The newly created cloned voice, identified by a unique voice ID returned by the Fish Audio API. This voice can be used for text-to-speech.	FISHAUDIO_VOICE

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): 6c4f011a4611a076b2488152591efeb61c029d6dfae2b079ba74689891c84803