Audio Concatenate

The opposite of Audio Cut is Audio Concatenate (SET_AudioConcatenate), and it does exactly what the name promises: takes two audio inputs and plays them back to back. Cut a track up into sections, then concatenate them in a different order and you've got yourself a primitive (and very ComfyUI-native) audio editor. It's also the node that makes multi-segment workflows reproducible - the "join" step that turns per-clip generation into a finished, ordered file.

How it works

The node takes audio1 and audio2 and outputs audio_out - audio1 followed immediately by audio2. Before it can glue two tensors together in time, though, it has to make them compatible, and it reuses the same alignment logic as Batch Audios:

• Sample rate: everything is resampled to audio1's rate. Input 1 is the reference; input 2 conforms to it.
• Channels: if one input is mono and the other stereo, the mono one is duplicated into "fake stereo" so both sides have two channels.
• Length: this is concatenation, so no padding to the longest - the output is simply the sum of both durations. Length matching only matters inside the batch dimension.

Batch handling is where it gets slightly clever: if the two inputs have different batch sizes, the last item of the shorter batch is repeated so the output batch matches the larger size. So a 2-item batch concatenated with a 5-item batch gives you a 5-item output where each of the first two clips is "item1 + itemA", "item2 + itemB", and the last three are "item2 + itemC/D/E" (last-of-short repeating). Not always what you want, but predictable.

How to think about it

Pair it with Audio Cut for reordering. Cut a long file into intro / verse / chorus, then concatenate them as chorus / verse / chorus and you've rearranged a song in the graph. Pair it with Audio Test Signal Generator and you can append a beep or a silent gap to the end of a clip - useful for pacing or just making sure your downstream sync has room to breathe.

Install & gotchas

From ComfyUI Manager, search "Audio Batch", or manually:

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

Restart ComfyUI. No model downloads.

The two things to watch: audio1's sample rate always wins, so put the "reference" file first if rate matters. And if your two clips are mono + stereo, the mono one becomes fake stereo - two identical copies of the same channel, not real stereo width. If you need true downmixing or actual stereo imaging, run a Channel Converter first. For getting the glue exactly where you want it, remember the durations are additive: a 2-second gap means feeding in a 2-second silence signal from the generator, not fiddling with a setting that doesn't exist here.