Skip to main content
Audio Split Export

Audio Split Export · screenshot coming soon

​
What this node does

Audio Split Export takes a single audio source and writes out multiple audio files, one per segment. Segments come from one of two modes: it can detect silence and split on the gaps, or take an explicit JSON list of start/end times. Each exported piece gets a short fade to avoid clicks, and the files are named by a chosen scheme. It accepts a COMFYTV_AUDIO snapshot or a COMFYTV_VIDEO (its audio track is used; audio takes priority). It outputs a COMFYTV_TEXT listing the exported files and has a ▶ Run (FFmpeg does the cutting). To feed native ComfyUI AUDIO in or out, insert a Bridge — see bridges.md.

​
When to use it

  • Break a long recording of separate phrases or sound effects into individual clips.
  • Auto-split a voice take into sentences by trimming on the pauses.
  • Export exact time ranges (e.g. from a cue sheet) as their own files.

​
Parameters

​
detect

How segment boundaries are found: silence (default) splits automatically on quiet gaps; json uses the ranges you provide in segments.

​
threshold_db

Loudness below which audio counts as silence, in dB. Range -90 to -20, default -60. silence mode only. Raise it (toward -20) if quiet passages are being treated as sound and not split.

​
min_silence_s

Minimum length a quiet gap must last to be treated as a split point, in seconds. Range 0.01 to 5, default 0.5. silence mode only.

​
min_segment_s

Minimum length a segment must have to be exported, in seconds. Range 0.01 to 10, default 0.1. Discards tiny slivers.

​
fade_ms

Fade in/out applied to each exported segment, in milliseconds. Range 0 to 500, default 1.45. A tiny fade prevents clicks at the cut edges.

​
naming

How exported files are named: num_and_prefix (default, number + the prefix), num_and_name (number + source name), or name (source name only).

​
prefix

The text used in the filename when naming is num_and_prefix. Default segment.

​
segments

A JSON list of ranges for json mode, in the form [{"start": s, "end": s}, ...] (times in seconds). Ignored in silence mode. Required (non-empty) when detect is json.

​
Outputs

​
Tips

  • In silence mode, if it splits too much, raise min_silence_s and/or lower threshold_db; if it doesn’t split enough, do the opposite.
  • Increase min_segment_s to throw away breaths and stray noises that would otherwise become their own tiny files.
  • json mode is exact and repeatable — use it when you already know the cut points.
  • Audio Crossfade — the inverse: join clips back together with a smooth blend.
  • Audio Mix — recombine segments into a layered mix.