ElevenLabs v4

Eleven v4 text to speech with a ready-made ElevenLabs voice: wire your text, pick a voice, run. Choose v4 (most natural), v4 Turbo (fastest) or v3. Add inline tags like [whispering], [laughing] or [shouting]. Returns audio plus word timestamps for captions.

Try ElevenLabs v4 in the playground

Last updated View as Markdown

ElevenLabs v4

Badge: v4
Node type: audio:elevenlabs_tts_eleven_v4
Category: audio

Description

ElevenLabs v4 reads your text aloud in one of ElevenLabs' ready-made voices: wire a Text Input, or any node that outputs text, into Text, pick a Voice, and click Run Model. It returns Audio, in an audio player that shows up on the node when the run finishes, plus Word Timestamps for captions. Choose v4 for the most natural model, v4 Turbo for speed, or v3, the previous expressive model, which also uses Style, Speed and Speaker Boost. On v4 and v4 Turbo you can steer the delivery with tags in square brackets, such as [whispering] or [laughing]. It costs 20 credits per run, flat, whatever the length of the text, and reads up to 2,500 characters; longer text is refused before anything is charged, so split a long script into parts. For a voice you cloned, use the ElevenLabs TTS (Custom Voice) node instead.

Pricing

  • Cost: ~20 credits per generation

Canvas ports

These appear as port handles on the left side of the node.

ID Label Details
text Text TEXT (required) — What to say, up to 2,500 characters per run. On v4 and v4 Turbo, add tags like [whispering] [laughing] [shouting].

These render as form fields in the right-side config panel when the node is selected.

ID Label Details
voice_id Voice TEXT (required) · options: pNInz6obpgDQGcFmaJgB, Xb7hH8MSUJpSbSDYk0k2, hpp4J3VqNfWAUOO0d1Us, pqHfZKP75CvOlQylNhV4, nPczCjzI2devNBz1zQrb, +16
model_id Model TEXT · options: eleven_v4, eleven_v4_turbo, eleven_v3 — v4 and v4 Turbo use only Stability and Similarity Boost. v3 also uses Style, Speed and Speaker Boost.
stability Stability NUMBER · 0…1
similarity_boost Similarity Boost NUMBER · 0…1
style Style NUMBER · 0…1
speed Speed NUMBER · 0.7…1.2
use_speaker_boost Speaker Boost BOOLEAN

Outputs

ID Label Type
audio Audio AUDIO
word_timestamps Word Timestamps JSON

How to use it

Wire a Text Input (or any node that outputs text) into Text, pick a voice, and click Run Model. No voice ID to paste: the Voice list is ElevenLabs' ready-made voices, and the search finds a voice by any part of its name, so "Lily" or "velvety" both find "Lily - Velvety Actress". The audio player shows up on the node when the run finishes.

Which model

  • v4 (the default) is ElevenLabs' most natural model.
  • v4 Turbo is the fastest.
  • v3 is the previous expressive model.

v4 and v4 Turbo read only Stability and Similarity Boost, so the sidebar shows only those two. Switch to v3 and Style, Speed and Speaker Boost come back.

What a run costs

20 credits per run, flat, whatever the length of the text. A run reads up to 2,500 characters. Longer text is refused before anything is charged, so split a long script into parts (one node per part, or an iterator) and join the audio.

Tags in the text

Write tags in square brackets right in the text to steer the delivery, for example [whispering] I have a secret [laughing] or [shouting] Go!.

Cloned voices

This node takes only ElevenLabs' ready-made voices; any other voice ID is refused before the charge. To speak in a voice you cloned, use ElevenLabs TTS (Custom Voice), which takes a voice ID or a wire from the ElevenLabs Voice Clone node and offers v4 too.


Auto-generated from the Wireflow node registry.

For AI agents: the full documentation index is at /llms.txt, and most docs pages are available as Markdown by adding .md to the URL.

© 2026 Wireflow. All rights reserved.