🎙️ Speechmatics LiveKit Agent

A simple LiveKit AI voice agent using Speechmatics for STT (Speech-to-Text) and TTS (Text-to-Speech).

---

⚡ Getting Up and Running

1. Install uv

We use uv for fast and reliable package management. Follow the installation guide here.

You can verify your installation by running:

uv --version

2. Install LiveKit CLI

We will be using the LiveKit command line interface to quickly get up and running.

🍎 macOS

brew install livekit-cli

🐧 Linux

curl -sSL [https://get.livekit.io/cli](https://get.livekit.io/cli) | bash

🪟 Windows

winget install LiveKit.LiveKitCLI

3. Link your LiveKit Cloud project to the CLI

(If running in console you can skip this step)

lk cloud auth

This will open a browser window allowing you to sign in or sign up to LiveKit Cloud and allowing your browser to connect to the agent we will create.

4. Initialise a project

uv init livekit-voice-agent --bare
cd livekit-voice-agent

5. Install packages

uv add \
  "livekit-agents[silero,turn-detector]~=1.2" \
  "livekit-plugins-noise-cancellation~=0.2" \
  "livekit-plugins-speechmatics" \
  "livekit-plugins-openai" \
  "python-dotenv" \
  "torch"

6. Get your API keys

Generate your local environment file:

lk app env > .env

This will create a .env.local file using the API keys and URLs linked to your LiveKit cloud account.

To this file, we will now add API keys for Speechmatics and OpenAI.

Note: Go to the Speechmatics website to sign up and get an API key.

Add these lines to your .env file:

SPEECHMATICS_API_KEY="<Your Speechmatics Key>"
OPENAI_API_KEY="<Your OpenAI Key>"

7. Create speechmatics_agent.py

Download, clone, or copy and paste the Python code into a file named speechmatics_agent.py in your project directory.

8. Download model files

Run the command to download models for voice activity detection and turn detection:

uv run speechmatics_agent.py download-files

9. Try your agent

Option A: Run in command line

uv run speechmatics_agent.py console

Option B: LiveKit Playground (UI) For a friendlier UI, connect it to the LiveKit Playground:

uv run speechmatics_agent.py dev