Comfy-Org/ComfyUI · error · ValueError

AudioSplit: Input audio must be stereo (2 channels), got {wa

Error message

AudioSplit: Input audio must be stereo (2 channels), got {waveform.shape[1]} channel(s).

What it means

The stereo splitter indexes waveform[..., 0:1, :] and [..., 1:2, :], so it requires exactly 2 channels in dim 1. Mono (1), 5.1 (6), or any other channel count raises immediately with the offending count.

Source

Thrown at comfy_extras/nodes_audio.py:510

            category="audio",
            inputs=[
                IO.Audio.Input("audio"),
            ],
            outputs=[
                IO.Audio.Output(display_name="left"),
                IO.Audio.Output(display_name="right"),
            ],
        )

    @classmethod
    def execute(cls, audio) -> IO.NodeOutput:
        if audio is None:
            return IO.NodeOutput(None, None)
        waveform = audio["waveform"]
        sample_rate = audio["sample_rate"]

        if waveform.shape[1] != 2:
            raise ValueError(f"AudioSplit: Input audio must be stereo (2 channels), got {waveform.shape[1]} channel(s).")

        left_channel = waveform[..., 0:1, :]
        right_channel = waveform[..., 1:2, :]

        return IO.NodeOutput({"waveform": left_channel, "sample_rate": sample_rate}, {"waveform": right_channel, "sample_rate": sample_rate})

    separate = execute  # TODO: remove

class JoinAudioChannels(IO.ComfyNode):
    @classmethod
    def define_schema(cls):
        return IO.Schema(
            node_id="JoinAudioChannels",
            display_name="Join Audio Channels",
            description="Joins left and right mono audio channels into a stereo audio.",
            category="audio",
            inputs=[
                IO.Audio.Input("audio_left"),

View on GitHub (pinned to 1c6d8d45b3)

Solutions

1. Check waveform.shape[1] before splitting and adapt (mono needs no split)
2. Convert to stereo first (torch.cat([wav, wav], dim=1)) if duplication is acceptable
3. Use SplitAudioChannels for arbitrary channel counts instead of the stereo-only AudioSplit
4. Fix the source/export to stereo if stereo is genuinely required

Example fix

// before
left, right = AudioSplit(audio)  # mono input -> raises

// after
if audio['waveform'].shape[1] == 1:
    audio = {'waveform': audio['waveform'].repeat(1, 2, 1), 'sample_rate': audio['sample_rate']}
left, right = AudioSplit(audio)

if waveform.shape[1] != 2:
    if waveform.shape[1] == 1:
        waveform = waveform.repeat(1, 2, 1)  # mono -> stereo
    else:
        waveform = waveform[:, :2, :]  # take first two channels

def is_stereo(audio) -> bool:
    return audio is not None and audio['waveform'].shape[1] == 2

When it happens

Trigger: Feeding mono audio (e.g. from JoinAudioChannel of a single channel, a mono TTS output, or a mono WAV via LoadAudio) into the split node; feeding multichannel surround audio.

Common situations: Workflows that assume line/video audio is stereo but receive mono narration; chains where an upstream channel op already collapsed to mono; device-specific mono capture.

Related errors

• AudioJoin: Both input audios must be mono.
• ERROR: audio encoder file is invalid or unsupported embed_di
• ERROR: audio encoder not supported.
• Image-token count {image_idx.shape[0]} != ViT output count {
• attention_mask must be {expected}, got {tuple(attention_mask

AI-assisted analysis of Comfy-Org/ComfyUI@1c6d8d45b3 (2026-08-14). Data as JSON: /api/errors/e09ff8a2b9335c19. Report an issue: GitHub.