Nodes/ComfyUI/ElevenLabs Voice Selector
ComfyUI Node Runs on cloud

ElevenLabs Voice Selector

Pick a voice from a dropdown — and it's actually free

By Comfy-Org·Created 4 years ago·Updated about 13 hours ago· 130,663
ElevenLabs Voice Selector
    • voice
    voice

    Rare honesty from a partner node: ElevenLabsVoiceSelector makes no API call and costs you nothing. It's a dropdown of 21 predefined ElevenLabs voices that exists purely to hand a voice down the wire. No credits, no bill, no server round-trip. In a family of paid nodes this one is free furniture - which makes it the natural starting point for any ElevenLabs workflow.

    The voices are the standard ElevenLabs cast, presented with name, gender, and accent: Roger, Sarah, Laura, Charlie (Australian), George and Daniel (British), River, Harry, Liam, Alice, Matilda, Will, Jessica, Eric, Bella, Chris, Brian, Lily, Adam, Bill, and a few more. Pick one; you'll hear the difference more than the label can tell you.

    How it fits

    The output is an ELEVENLABS_VOICE - a handle, not audio - and it feeds the nodes that take a voice as a wire input: ElevenLabsTextToSpeech, ElevenLabsSpeechToSpeech, and the per-line voices in ElevenLabsTextToDialogue. This is the recurring pattern in the ElevenLabs group that trips people up: those nodes don't have a voice dropdown of their own. The dropdown lives here, and you connect it with a wire. Forgetting to wire one in is the single most common "why won't this run" moment in the whole family.

    The one input, the one output

    • voice - the dropdown. That's the entire interface.
    • Output: voice, type ELEVENLABS_VOICE.

    The honest notes

    • It's free, but it's also just a catalog. If you want your voice, that's ElevenLabsInstantVoiceClone, which costs a flat fee per clone and produces the same ELEVENLABS_VOICE output - both feed the same inputs.
    • The predefined voices can change as ElevenLabs rotates its lineup. A workflow saved against a specific name may want a re-pick after a while.
    • One selector can feed many speech nodes, but the same voice wire can't be two different voices - for multi-speaker work, add a selector (or a clone) per character.

    Ships with ComfyUI core (landed February 2026); nothing to install. It's the most boring node in the group, and that's the compliment - it's the bit that makes every other ElevenLabs node usable.

    Categorypartner/audio/ElevenLabs

    Inputs (1)

    NameTypeDefaultDescription
    voiceCOMBOChoose a voice from the predefined ElevenLabs voices.

    Outputs (1)

    NameTypeDescription
    voiceELEVENLABS_VOICE