Comfy-Org/ComfyUI · error · ValueError

Too many reference audios: {total_audios} (audios={len(refer

Error message

Too many reference audios: {total_audios} (audios={len(reference_audios)}, audio assets={len(reference_audio_assets)}). Maximum is {limits['max_audios']}.

What it means

Client-side cap on reference audios (Seedance 2.5 feature): direct reference_audios plus resolved audio assets must not exceed limits["max_audios"] (10). Audio references are used for voice/timbre transfer, and the provider rejects more than the cap, so the node enforces it locally with a breakdown of both sources.

Source

Thrown at comfy_api_nodes/nodes_bytedance.py:2766

                raise ValueError("At least one reference image or video or asset is required.")

        total_images = len(reference_images) + len(reference_image_assets)
        if total_images > limits["max_images"]:
            raise ValueError(
                f"Too many reference images: {total_images} "
                f"(images={len(reference_images)}, image assets={len(reference_image_assets)}). "
                f"Maximum is {limits['max_images']}."
            )
        total_videos = len(reference_videos) + len(reference_video_assets)
        if total_videos > limits["max_videos"]:
            raise ValueError(
                f"Too many reference videos: {total_videos} "
                f"(videos={len(reference_videos)}, video assets={len(reference_video_assets)}). "
                f"Maximum is {limits['max_videos']}."
            )
        total_audios = len(reference_audios) + len(reference_audio_assets)
        if total_audios > limits["max_audios"]:
            raise ValueError(
                f"Too many reference audios: {total_audios} "
                f"(audios={len(reference_audios)}, audio assets={len(reference_audio_assets)}). "
                f"Maximum is {limits['max_audios']}."
            )

        for key in reference_images:
            reference_images[key] = _prepare_seedance_image(reference_images[key])

        has_video_input = total_videos > 0

        if model.get("auto_downscale") and reference_videos:
            max_px = SEEDANCE2_REF_VIDEO_PIXEL_LIMITS.get(model_id, {}).get(model["resolution"], {}).get("max")
            if max_px:
                for key in reference_videos:
                    reference_videos[key] = downscale_video_to_max_pixels(reference_videos[key], max_px)

        if model.get("auto_upscale") and reference_videos:
            min_px = SEEDANCE2_REF_VIDEO_PIXEL_LIMITS.get(model_id, {}).get(model["resolution"], {}).get("min")

View on GitHub (pinned to 1c6d8d45b3)

Solutions

Reduce the number of audio references (direct + assets) to the stated maximum.
Pick the one or two cleanest voice samples; more samples rarely improve timbre.

Defensive patterns

Strategy: validation

Validate before calling

total = len(reference_audios) + len(reference_audio_assets)
assert total <= seedance2_reference_limits(model_id)["max_audios"]

Prevention

One or two clean voice samples are usually enough for timbre transfer.
Count clips and audio asset ids together against the cap.

When it happens

Trigger: Connecting more reference_audio inputs than max_audios, or combining reference_audios with audio asset ids above the cap, on the Seedance 2.5 reference node.

Common situations: Trying to blend many voice samples; mixing preset workflows that carry asset ids with newly connected audio clips.

Related errors

AI-assisted analysis of Comfy-Org/ComfyUI@1c6d8d45b3 (2026-08-14). Data as JSON: /api/errors/c735e58f8027d9d7. Report an issue: GitHub.