Nodes/ComfyUI-kaola-ace-step/ACE-Step Captioner
ComfyUI Node

ACE-Step Captioner

A ComfyUI node in Audio/ACE-Step with 12 inputs and 3 outputs.

By kana112233·Created 6 months ago·Updated 6 months ago· 27
ACE-Step Captioner
  • audio
  • caption
  • style_tags
  • full_description
model_idACE-Step/acestep-captioner
deviceauto
dtypeauto
custom_prompt*Task* Describe this audio in detail
max_new_tokens1024
temperature0.3
top_p0.90
top_k50
repetition_penalty1.1
seed0
chunk_length_s30
CategoryAudio/ACE-Step

Inputs (12)

NameTypeDefaultDescription
audioAUDIOInput audio to caption/describe.
model_idCOMBOACE-Step/acestep-captionerSelect the captioner model. Can be a local path or HuggingFace ID.
deviceCOMBOautoInference device. Use 'auto' or 'mps' for Mac.
dtypeCOMBOautoModel precision. 'auto' uses float16 for CUDA and float32 for CPU/MPS.
custom_promptoptSTRING*Task* Describe this audio in detailCustom prompt for captioning. Default is the recommended prompt from ACE-Step.
max_new_tokensoptINT102464–4096Maximum number of tokens to generate. Increase for longer descriptions.
temperatureoptFLOAT0.30–1Sampling temperature. Lower values (0.1-0.3) are more deterministic and accurate.
top_poptFLOAT0.900–1Nucleus sampling: cumulative probability threshold.
top_koptINT500–1000Top-K sampling. 0 = disabled. Lower values can improve accuracy.
repetition_penaltyoptFLOAT1.11–2Penalty for repeating tokens. Increase if output gets stuck in loops.
seedoptINT00–4294967295Random seed for reproducible results. 0 for random.
chunk_length_soptFLOAT300–300Audio chunk length in seconds for processing long audio.

Outputs (3)

NameTypeDescription
captionSTRING
style_tagsSTRING
full_descriptionSTRING