MiniMaxMusic3TextEncode - ComfyUI Built-in Node Documentation - ComfyUI

MiniMax Music3 Text Encode uses a MiniMax Music3 CLIP model to convert text captions and lyrics into an acoustic conditioning sequence for music generation. The node returns the resulting CONDITIONING data, along with the actual audio duration in seconds calculated from the input maximum duration.

Inputs

Parameter	Description	Data Type	Required	Range
`clip`	The MiniMax Music3 CLIP model, used for text encoding and conditioning sequence generation.	CLIP	Yes	-
`caption`	Text describing the music to generate. Supports multi-line text and dynamic prompts.	STRING	Yes	-
`lyrics`	The lyrics text to be used for generating the music. Supports multi-line text and dynamic prompts.	STRING	Yes	-
`seed`	Reproducible random seed for the generation process. Default: 0.	INT	Yes	0 to 18446744073709551615 (0xffffffffffffffff)
`max_duration`	Maximum duration in seconds; the model can end the song earlier. Default: 120.0.	FLOAT	Yes	0.04 to the model’s maximum audio duration (MAX_AUDIO_FRAMES / AUDIO_FRAMES_PER_SECOND), step 0.04
`cfg_scale`	Classifier-free guidance scale. Default: model constant CFG_SCALE. Advanced parameter.	FLOAT	Yes	0.0 to 100.0, step 0.1 (keeps 2 decimal places)
`top_k`	Top-k sampling value used for acoustic token selection. Default: model constant CFG_TOP_K. Advanced parameter.	INT	Yes	1 to the model’s vocabulary size (C0_VOCAB_SIZE)

Outputs

Output Name	Description	Data Type
`conditioning`	The generated acoustic conditioning sequence, used to guide subsequent music generation.	CONDITIONING
`seconds`	The actual duration of the conditioning sequence, in seconds.	FLOAT

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): c3fbfd189d0358ebf081dd4f9c32be9231a9d0b97fd767401ea4b7955224c25c