Nodes/ComfyUI-Qwen3-ASR/Qwen3 ASR Transcriber
ComfyUI Node

Qwen3 ASR Transcriber

A ComfyUI node in Qwen3-ASR with 10 inputs and 2 outputs.

By kaushiknishchay·Created 6 months ago·Updated 5 months ago· 25
Qwen3 ASR Transcriber
  • audio
  • forced_aligner
  • text
  • timestamps
model_name
languageauto
devicecuda
precisionbf16
max_new_tokens256
flash_attention_2false
chunk_size30
overlap2
CategoryQwen3-ASR

Inputs (10)

NameTypeDefaultDescription
audioAUDIOThe input audio to be transcribed.
model_nameCOMBOThe Qwen3 ASR model to use for transcription.
languageCOMBOautoThe language of the audio. Set to 'auto' for automatic language detection.
deviceCOMBOcudaThe device to run the ASR model on.
precisionCOMBObf16The numerical precision to use for the ASR model.
max_new_tokensINT2561–4096The maximum number of tokens to generate in the transcription.
flash_attention_2BOOLEANfalseEnable Flash Attention 2 for faster inference and lower VRAM usage.
chunk_sizeINT300–300Process audio in chunks of this many seconds. Set to 0 to disable chunking (not recommended for long audio).
overlapINT20–10Overlap between chunks in seconds to maintain context.
forced_aligneroptQWEN3_ALIGNER_CONFOptional configuration for the Qwen3 Forced Aligner to generate word-level timestamps.

Outputs (2)

NameTypeDescription
textSTRING
timestampsSTRING