Audio Force Channels
Audio Force Channels — when a number is easier than a dropdown
- audio
- audio
Audio Force Channels (SET_AudioForceChannels) is the integer-flavored sibling of Audio Channel Converter. Where the Converter gives you a dropdown of named strategies, this node asks for one number: channels. And because it's an integer, it slots straight into anything that computes a value - a fixed 2 for your video pipeline, a 1 for a voice-only branch, whatever logic upstream decides.
The mapping is the entire node, and it's clean:
0→ keep as-is (the Converter'skeep).1→ force mono.2→ force stereo.
That's the full range (min 0, max 2). There's no 3 or 8 - the node is deliberately a mono/stereo gate, and it says so in the tooltip: "Forces the number of channels. 0 means keep same."
Everything else is shared with its sibling. The downmix_method dropdown (average, standard_gain default, spectral) governs how multi-channel audio collapses to mono, and the optional n_fft / hop_length inputs tune the spectral method when you pick it. If you've read the Channel Converter article, none of this is new - it's the same machinery behind a numeric interface.
Why the numeric version exists
Honestly? Convenience and automation. In a hand-built workflow, the dropdown is nicer to read. But the moment your channel count is driven - a preset switcher, a batch of workflow variants, a value you want to expose as a widget on the front of the graph - a number beats a dropdown because you can compute it. Also worth noting: with channels = 0, this is a harmless passthrough that just logs a warning on exotic channel counts, which makes it a decent place to stash an audio branch you might convert later.
Output
audio_out - the converted audio, sample rate and batch size untouched. It changes channels and only channels, just like the Converter.
Install & gotchas
Part of 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 one trap is forgetting what 0 means. Because 0 is "keep," the default state does nothing - a workflow that passes audio through this node with channels left at 0 will look like it's converting when it isn't. If you're debugging why a downstream node still sees mono, check this first. And remember the same fake-stereo caveat as its sibling: forcing to 2 channels duplicates the mono channel; it does not invent stereo width.
Inputs (5)
| Name | Type | Default | Description |
|---|---|---|---|
| audio | AUDIO | — | |
| channels | INT | 00–2 | — |
| downmix_method | COMBO | standard_gain | Method for stereo/multi-channel to mono conversion: - average: Simple average ((L+R)/2). Can reduce volume. - standard_gain: Sums channels with -3 dB gain (0.707). Better preserves perceived loudness.- spectral: Averages frequency magnitudes to prevent phase cancellation. |
| n_fftopt | INT | 2048256–8192 | FFT size for spectral downmixing. Higher values give better frequency resolution but worse time resolution. |
| hop_lengthopt | INT | 51264–4096 | Hop length for STFT. Typically n_fft / 4. Controls time resolution. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| audio | AUDIO | — |