Text-to-Speech

Overview

soniox.TTS provides real-time text-to-speech synthesis using the Soniox WebSocket TTS API. It delivers:

Ultra-low latency — synthesis starts from the first few words, before the full sentence is available, keeping voice agents responsive.
60+ languages with native-speaker-quality voices. See supported languages and voices.
Accurate alphanumerics — phone numbers, email addresses, IDs, and other mixed letter-and-digit strings are pronounced correctly.
Hallucination-free output — the model speaks the text you send, nothing more.

Basic usage

Use Soniox TTS in an AgentSession:

from livekit.agents import AgentSession
from livekit.plugins import soniox

session = AgentSession(
    tts=soniox.TTS(model="tts-rt-v1", voice="Maya"),
    # ... stt, llm, etc.
)

Configuration

soniox.TTS accepts top-level constructor arguments for the model, voice, language, and audio settings.

Constructor arguments

Argument	Type	Default	Description
`api_key`	`str \| None`	`SONIOX_API_KEY` env var	Soniox API key.
`model`	`str`	`tts-rt-v1-preview`	TTS model identifier.
`language`	`str`	`en`	Language code (e.g., `"en"`, `"es"`, `"fr"`). See supported languages.
`voice`	`str`	`Maya`	Voice identifier. See voices.
`audio_format`	`str`	`pcm_s16le`	Output audio format. See audio formats.
`sample_rate`	`int`	`24000`	Output sample rate in Hz.
`bitrate`	`int \| None`	`None`	Codec bitrate in bps for lossy compressed formats.
`websocket_url`	`str`	`wss://tts-rt.soniox.com/tts-websocket`	Soniox WebSocket endpoint. See regional endpoints.
`http_session`	`aiohttp.ClientSession \| None`	`None`	Optional aiohttp session to reuse for the WebSocket connection.

Advanced usage

Regional endpoints

If you want to use a different region than the default (US), pass a regional websocket_url to the TTS constructor:

from livekit.plugins import soniox

tts = soniox.TTS(websocket_url="wss://tts-rt.eu.soniox.com/tts-websocket")

See the list of regional endpoints for available endpoints.

Voice and language

Soniox TTS supports 60+ languages and a range of voices. Specify them via the constructor:

from livekit.plugins import soniox

tts = soniox.TTS(
    model="tts-rt-v1",
    voice="Maya",
    language="en",
)

See the list of supported languages and available voices.

Updating options at runtime

Use update_options() to change the model, language, or voice mid-session without recreating the TTS instance:

tts.update_options(voice="Adrian", language="es")