Nodes/KIE.ai Nodes Next/Suno Voice Regenerate Verification Phrase
ComfyUI Node

Suno Voice Regenerate Verification Phrase

When your one good take is gone

By felipederosilva·Created about a month ago·Updated a day ago· 3
Suno Voice Regenerate Verification Phrase
    • audio
    • url
    • all_urls_json
    • task_id
    • raw_json
    • credits_consumed
    • credits_left
    ◄task_idxxxxxxx►
    ◄timeout_seconds1200►
    ◄callback_url►

    Voice cloning is a two-take job by nature: you record a phrase, and half the time you hate the take. This node is the reset button - it asks KIE for a fresh verification phrase off an existing voice task, so you get another shot without rebuilding the pipeline from scratch.

    Where it sits in the chain

    The Suno voice flow on KIE is: generate a verification phrase from a slice of your recording, record yourself performing it, submit that recording to create the custom voice, then check when the voice is ready. This node hangs off the second step - you already have a voice task, and you want a new phrase for it.

    Reach for it in three cases. The first is the obvious one: you fluffed the take. The second is when KIE's returned phrase expired or the audio link you saved stopped resolving - these endpoints hand out temporary URLs, so a phrase you bookmarked last month is probably gone. The third is when the phrase itself is awkward: an odd rhythm or a range you can't comfortably hit will produce a worse clone, and a different phrase can genuinely save the result. Singing well matters more here than in almost any other API call you'll make - the provider is assessing your voice from this performance.

    Inputs and outputs

    There's exactly one field you must fill: task_id, the ID of the voice task you want a new phrase for. The widget's default is a placeholder (xxxxxxx), so paste the real one - or, better, wire the task_id output from the node that produced it and stop babysitting strings.

    timeout_seconds (30–7200 seconds, default 1200) is the completion window. callback_url should be left empty for normal ComfyUI use; it's there for servers that want KIE to POST back instead of being polled.

    Outputs follow the pack's standard audio signature: audio, url, all_urls_json, task_id, raw_json, credits_consumed, credits_left. In practice you read the phrase out of raw_json, play the audio output through a core Save Audio or preview node to hear what you're supposed to sing, and carry the returned task_id forward into the create-voice node. credits_consumed and credits_left are there if you like watching the meter - worth it, because a few rounds of "wrong phrase, try again" adds up on a pay-as-you-go account.

    Install

    Shared setup for every node in the pack - ComfyUI Manager, search KIE.ai Nodes Next, or by hand:

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

    Restart ComfyUI afterwards. Then open Settings → KIE.ai Nodes Next → Connection and paste your API key once; the pack checks it against KIE before saving, shows you a masked fingerprint, and reports CONNECTED along with your credit balance. On a headless box, set KIE_API_KEY in the environment instead - it takes priority over the saved key. Dependencies are requests, Pillow, numpy and PyYAML, and there are no model files to download, because all the work happens on KIE's GPUs.

    Things that go wrong

    Passing the wrong ID is the classic. task_id here means the voice task, not the verification-phrase task you may still have in the graph - they're both "the task" in casual conversation and they are not interchangeable.

    Then there's the category-level caveat: this pack builds its node list from KIE's live documentation and re-checks at most daily, so a voice endpoint KIE renames, retires or reshapes will show up in your graph looking fine and then fail. When the inputs no longer line up with what the endpoint documents, run a catalog sync and re-read the node rather than assuming your workflow rotted.

    And the one that isn't a bug: the returned phrase links are temporary by design. Save the audio locally if you're going to come back to a take later, rather than assuming the URL still works next week.

    One security note, since every node here holds a credential and calls the network as its job: prefer packs with real community mileage, and skim what they do with your key. That's true of this one, of the older ComfyUI-Kie-API pack, and of the official ComfyUI API nodes alike.

    CategoryKIE Next/Audio/Suno

    Inputs (3)

    NameTypeDefaultDescription
    task_idSTRINGxxxxxxxTask ID
    timeout_secondsoptINT120030–7200—
    callback_urloptSTRINGOptional KIE callback URL. Leave empty for normal ComfyUI use.

    Outputs (7)

    NameTypeDescription
    audioAUDIO—
    urlSTRING—
    all_urls_jsonSTRING—
    task_idSTRING—
    raw_jsonSTRING—
    credits_consumedFLOAT—
    credits_leftFLOAT—