WanDancerEncodeAudio - ComfyUI Built-in Node Documentation

This node analyzes an audio clip and turns it into a set of features that can guide a video generation model.

This node analyzes an audio clip and turns it into a set of features that can guide a video generation model. It estimates tempo and beats, extracts mel-spectrogram, MFCC, chroma, and onset features, then packages them together with a calculated frame rate for synchronization.

Inputs

Parameter	Description	Data Type	Required	Range
audio	The audio input to be analyzed and encoded. If the audio has multiple channels, the channels are averaged to mono before feature extraction.	AUDIO	Yes	-
video_frames	The number of frames in the target video. Used to calculate the frame rate for synchronization (default: 149).	INT	Yes	Min: 1, Max: 16384 (MAX_RESOLUTION), Step: 4
audio_inject_scale	The scale for the audio features when injected into the video model (default: 1.0).	FLOAT	Yes	Min: 0.0, Max: 10.0, Step: 0.01

Outputs

Output Name	Description	Data Type
audio_encoder_output	A dictionary containing the processed audio features, the calculated frame rate (fps), and the audio injection scale. This output is used to condition the video generation model.	AUDIO_ENCODER_OUTPUT
fps_string	A text string describing the calculated frame rate (fps) based on the audio length and the number of video frames. This string is intended to be used in the prompt for the video model. It is formatted in Chinese to match the reference pipeline.	STRING

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): ce27a3bdea2d9e3cf8875c24236a2a0a1429e9bc13a58581e372fb669d2c0018