Text to speech

Turns text into spoken audio in the voice you pick.

Step by step

1. Add a Generate audio node and set Operation to Text to speech.
2. Write the script in the node, or wire a text node into it. Example: Meet the new runner. Lighter, faster, made to move.
3. Pick a Model: eleven_multilingual_v2 (default), eleven_v3, eleven_flash_v2_5 or eleven_turbo_v2_5.
4. Pick a Voice. Standard voices are the default ElevenLabs voices; My voices are your organization’s own (see Voices). The play button previews the selected voice. With nothing picked, the voice is George.
5. Optionally move Stability, Similarity and Style (0 to 1). They are sent only once you touch them.
6. Press Generate. The MP3 lands in the node’s audio output.

Ports

Input: text. Output: audio.

Cost

Per character of the script: $0.08 per 1,000 on eleven_multilingual_v2 and eleven_v3, $0.04 per 1,000 on the flash and turbo models. See Audio.

Agents and MCP

run_node_generation with medium: "audio":

{ "node_id": "<audio node>", "medium": "audio", "version": 3,
  "params": { "operation": "text_to_speech", "voiceId": "<voice id>", "stability": 0.5 } }

The script is the node’s prompt or connected text. describe_node_types with type: audio lists voice ids.

A voice that belongs to another workspace is refused (voice_not_yours), and cloned voices cannot be used in uncensored projects.