Nodes/ComfyUI_AIIA/AIIA Generate Speaker Segments
ComfyUI Node

AIIA Generate Speaker Segments

A ComfyUI node in AIIA/audio with 11 inputs and 1 output.

By havvk·Created about a year ago·Updated 6 months ago· 12
AIIA Generate Speaker Segments
  • audio
  • speaker_segments
e2e_backend_modelNO_MODELS_FOUND
num_speakers_hint0
postprocessing_profilebalanced
devicecuda
custom_onset0.60
custom_offset0.40
custom_min_duration_on0.10
custom_min_duration_off0.20
custom_pad_onset0.05
custom_pad_offset0.05
CategoryAIIA/audio

Inputs (11)

NameTypeDefaultDescription
audioAUDIO
e2e_backend_modelCOMBONO_MODELS_FOUND1 options: NO_MODELS_FOUND
num_speakers_hintINT00–4期望说话人数 (0=模型自动估计, 最多4人)。此为提示。
postprocessing_profileCOMBObalanced选择后处理参数配置方案。
deviceoptCOMBOcuda2 options: cuda, cpu
custom_onsetoptFLOAT0.600–1(Custom Profile) VAD onset threshold
custom_offsetoptFLOAT0.400–1(Custom Profile) VAD offset threshold
custom_min_duration_onoptFLOAT0.100.01–2(Custom Profile) Min duration for a speech segment
custom_min_duration_offoptFLOAT0.200.01–2(Custom Profile) Min duration for a non-speech segment
custom_pad_onsetoptFLOAT0.05-0.5–0.5(Custom Profile) Padding for speech segment onset
custom_pad_offsetoptFLOAT0.05-0.5–0.5(Custom Profile) Padding for speech segment offset

Outputs (1)

NameTypeDescription
speaker_segmentsWHISPER_CHUNKS