Soniox plugin for LiveKit Agents

Support for Soniox Speech-to-Text and Text-to-Speech APIs, using WebSocket streaming interfaces.

See STT documentation and TTS documentation for more information.

Installation

pip install livekit-plugins-soniox

Pre-requisites

The Soniox plugin requires an API key to authenticate. You can get your Soniox API key here.

Set API key in your .env file:

SONIOX_API_KEY=<your_soniox_api_key>

Usage

Speech-to-Text (STT)

Use Soniox STT in an AgentSession or as a standalone transcription service:

from livekit.plugins import soniox

session = AgentSession(
    stt=soniox.STT(),
    # ... llm, tts, etc.
)

Text-to-Speech (TTS)

Use Soniox TTS for real-time speech synthesis:

from livekit.plugins import soniox

session = AgentSession(
    tts=soniox.TTS(
        language="en",
        voice="Maya",
        speed=1.0,  # 0.7-1.3
    ),
    # ... stt, llm, etc.
)

The TTS supports real-time streaming from LLM - text chunks are tokenized and sent to Soniox as words are formed, enabling low-latency speech synthesis.

More information and reference

Explore integration details and find comprehensive examples: