ElevenLabs Pro - Text to Speech

A ComfyUI node in ElevenLabs Pro/TTS with 18 inputs and 2 outputs.

By IxMxAMAR·Created 4 months ago·Updated 2 months ago· 1

audio
raw_response_info

◄api_key►

◄text►

◄voice_id►

◄modeleleven_v3►

◄stability0.50►

◄similarity_boost0.75►

◄style0.00►

◄speed1.00►

◄use_speaker_boosttrue►

◄languageAuto Detect►

◄apply_text_normalizationauto►

◄output_formatmp3_44100_128►

◄seed0►

◄previous_text►

◄next_text►

◄use_pvc_as_ivcfalse►

◄enable_loggingtrue►

◄pronunciation_dictionary_locators►

CategoryElevenLabs Pro/TTS

Inputs (18)

Name	Type	Default	Description
api_key	STRING		—
text	STRING		Text to convert to speech. Supports v3 audio tags like [whispers], [laughs], [excited].
voice_id	STRING		Voice ID. Connect from Voice Selector or Fetch Voices node.
model	COMBO	eleven_v3	8 options: eleven_v3, eleven_multilingual_v2, eleven_flash_v2_5, eleven_turbo_v2_5, eleven_flash_v2, eleven_turbo_v2, +2
stabilityopt	FLOAT	0.500–1	Voice stability. Lower = more expressive/emotional, Higher = more consistent/monotone. Creative(<0.5), Natural(0.5), Robust(>0.5).
similarity_boostopt	FLOAT	0.750–1	How closely AI adheres to the original voice. Too high may introduce artifacts.
styleopt	FLOAT	0.000–1	Style exaggeration. Increases expressiveness but reduces stability. Recommended: 0 for most use cases.
speedopt	FLOAT	1.000.5–2	Speech speed. 1.0 = normal, <1.0 = slower, >1.0 = faster.
use_speaker_boostopt	BOOLEAN	true	Boost similarity to original speaker. Increases latency slightly.
languageopt	COMBO	Auto Detect	Language for the model. Auto Detect lets the model decide. NOTE: turbo/flash v2.5 models require an explicit language.
apply_text_normalizationopt	COMBO	auto	Text normalization mode. eleven_v3 is auto-forced to 'off' (the only valid value for that model).
output_formatopt	COMBO	mp3_44100_128	Audio output format. mp3_44100_192 and opus require Creator tier+.
seedopt	INT	00–4294967295	Seed for reproducibility. 0 = random. Determinism not guaranteed.
previous_textopt	STRING		Context only — text that came BEFORE this segment. Does NOT generate audio for this text. Used for seamless chunking across long inputs.
next_textopt	STRING		Context only — text that comes AFTER this segment. Does NOT generate audio for this text. Used so the model can anticipate pacing.
use_pvc_as_ivcopt	BOOLEAN	false	Use the IVC version of a Professional Voice Clone.
enable_loggingopt	BOOLEAN	true	If False, requests zero-retention mode (audio + text not stored by ElevenLabs). Required for HIPAA / privacy-sensitive content.
pronunciation_dictionary_locatorsopt	STRING		JSON array of {"pronunciation_dictionary_id": ..., "version_id": ...} objects.

Outputs (2)

Name	Type	Description
audio	AUDIO	—
raw_response_info	STRING	—