Audio Cut

Every audio workflow eventually needs less audio than it started with. Maybe a loaded track has a long silent intro you want gone, or you only need the first 15 seconds of a generated clip. Audio Cut (SET_AudioCut) slices any window out of a clip, and it's the direct counterpart to Audio Concatenate - cut to trim, concatenate to rebuild. It's also the one node in this pack with famous DNA: the README credits it as highly based on christian-byrne's audio-separation-nodes-comfyui, which is why the time parsing here is friendlier than most.

The inputs

• audio - whatever you're cutting.
• start_time - where the cut begins.
• end_time - where it ends.

The trick that makes this node nicer than most: both time fields are strings that accept either a plain float in seconds or an HH:MM:SS.ss-style time. So 1:30 means one minute thirty seconds, while 90.5 means the same thing written as seconds. You can use 0:00 to 1:00 to grab the first minute, or 12.5 to 30 for a raw-samples-free seconds window. No unit dropdowns, no milliseconds math.

The output, audio_out, is just that window - same sample rate, same channels, only shorter. The rest of the file is discarded, not hidden.

How you'll actually use it

Trim is the obvious one: drop a long track, cut out the dead space. But it really shines in assembly workflows with Audio Concatenate. Cut a file into sections, reorder them, concatenate back - you've rearranged audio without ever touching an external editor, and the whole thing is reproducible from the workflow JSON. Loop it with a couple of fixed windows and you can build a primitive sampler: pick a slice, repeat it via concatenation.

One thing to know: the cut is per-batch-item and works on batches, so you can carve the same window out of a whole batch of clips in one go - handy when every clip in a batch needs the same leading silence removed.

Install & gotchas

It ships with the Audio Batch pack. ComfyUI Manager, search "Audio Batch", or:

cd ComfyUI/custom_nodes/
git clone https://github.com/set-soft/ComfyUI-AudioBatch
pip install seconohe

Restart ComfyUI. No model downloads.

The gotcha is the string format. 1:30 is parsed as one minute thirty - but if you're in the habit of typing 1.5 expecting 1.5 seconds, that's fine (it's a float, 1.5 seconds). The ambiguity only bites if you mix conventions: 1.30 is not a time, it's the float 1.3 seconds. And if your end time lands past the end of the clip, expect the cut to clamp rather than extend - the window just ends where the audio does.