Upload And Cover Audio · V5

A cover is one of the few music jobs where an API genuinely has no local equal. You hand KIE a track, describe the style, and get a new performance back - with the vocal quality that is the whole reason people still pay for Suno instead of running ACE-Step. This node is that endpoint at the V5 tier, which KIE's docs mark as discontinued.

That marker doesn't make it useless. It means: for new work, use the V6 sibling; for re-running a cover you made months ago, or for a workflow already pinned to V5, this is the node that keeps the output identical. Covers are also cheap enough per call that the tier question is mostly about sound, not cost.

How it works

The pack generates one node class per documented model version from KIE's API docs at startup, so the tier is baked into the class - no model dropdown, nothing to accidentally change. The node assembles the documented request body, submits it to KIE's task endpoint (envelope model ai-music-api/upload-and-cover-audio, inner model V5), polls the record-info endpoint every couple of seconds with exponential backoff until the task completes, then downloads the resulting audio and returns a native ComfyUI AUDIO. Errors raise with the task id and the provider's message rather than silently returning nothing.

The inputs that matter

Three required:

• upload_url - the source track, as a URL KIE can fetch. The docs cap source audio at 8 minutes and say the source alone is enough to generate a result.
• prompt - the docs are specific: it's used as lyrics only when lyrics isn't provided. Set lyrics and the prompt stops being lyrics. And when instrumental is true, don't pass it at all.
• instrumental - default true, i.e. no vocals. That default surprises people who came here to make a sung cover; flip it to false.

Optional: style (the genre - up to 1000 characters on this tier), title (80 characters), negative_tags, vocal_gender (m/f; the docs only claim it shifts the odds), style_weight, weirdness_constraint, audio_weight, lyrics (takes priority over prompt), variety (0–4; 0 pins the style, 4 wanders), duration (10–360 seconds - note the tooltip says it applies to V5.5/V6 and not this tier), persona_id and persona_model (Suno's reusable persona feature, which the docs scope to V5, V5.5 and V6 explicitly - here it's actually sanctioned), plus timeout_seconds and callback_url.

Outputs: audio, url, all_urls_json (the response carries both the source and the cover URL - this is where the full set lives), task_id, raw_json, credits_consumed, credits_left.

Feeding it a file

upload_url is a text box, so there's no dragging an AUDIO onto it. Add KIE • Upload Audio from KIE Next/Utility/Media, feed it the audio you want covered, then right-click upload_url on this node, pick Convert widget to input, and connect the returned URL string. Two extra nodes and you never paste a URL again.

Installing

ComfyUI Manager, search "KIE Nodes Next". The README describes registry publication as prepared rather than finished, so if Manager shows nothing, install manually:

cd ComfyUI/custom_nodes
git clone https://github.com/felipederosilva/ComfyUI-KIE-Nodes-Next
pip install -r ComfyUI-KIE-Nodes-Next/requirements.txt

Restart, then paste your key once at Settings → KIE.ai Nodes Next → Connection. It's validated before it overwrites a working key, the UI only ever sees a masked fingerprint, and the key never lands in workflow JSON - which matters, because API-node packs are exactly the category where a leaked key is the failure mode. KIE_API_KEY beats the saved key on headless installs. Dependencies: four small libraries, no torch.

Where people get burned

• Instrumental is on by default. Fill in lyrics, run it, get a karaoke track - that's the default, not a bug.
• Prompt vs. lyrics precedence. If your words aren't being sung, check whether lyrics is set; it wins over prompt every time.
• No key configured gives an explicit error naming the settings page.
• Mixed-release installs are refused at import time. If the pack says its version and generated-node build disagree, delete the folder and reinstall.
• Not a legal opinion. Covering someone else's song is a licensing question wherever you publish it; the node doesn't care and neither does the model.
• V5 is discontinued per KIE's docs - expect to move to the V6 node eventually.

If you're covering for a video, the workflow that holds up is: get the cover right first, then feed its audio output into whatever makes the video. Chasing both at once means you can't tell which change made the mouth stop matching.