You will need an EchoKit device, or create your own ESP32 device with the EchoKit firmware.
git clone https://github.com/second-state/echokit_server
Edit config.toml to customize the VAD, ASR, LLM, TTS services, as well as prompts and MCP servers. You can see many examples.
cargo build --release
The config.toml can use any combination of open-source or proprietary AI services, as long as they offer OpenAI-compatible API endpoints. Here are instructions to start open source AI servers for the EchoKit server.
- VAD: https://github.com/second-state/silero_vad_server
- ASR: https://llamaedge.com/docs/ai-models/speech-to-text/quick-start-whisper
- LLM: https://llamaedge.com/docs/ai-models/llm/quick-start-llm
- Streaming TTS: https://github.com/second-state/gsv_tts
Alternatively, you could use Google Gemini Live services for VAD + ASR + LLM, and even optionally, TTS. See config.toml examples.
You can also configure MCP servers to give the EchoKit server tool use capabilities.
The hello.wav file on the server is sent to the EchoKit device when it connects. It is the voice prompt the device will say to tell the user that it is ready.
export RUST_LOG=debug
nohup target/release/echokit_server &
Go here: https://echokit.dev/chat/
Click on the link to save the index.html file to your local hard disk.
Double click the local index.html file and open it in your browser.
In the web page, set the URL to your own EchoKit server address, and start chatting!
Go to web page: https://echokit.dev/setup/ and use Bluetooth to connect to the GAIA ESP332 device.
Configure WiFi and server
- WiFi SSID (e.g.,
MyHome) - WiFi password (e.g.,
MyPassword) - Web Socket server URL for
echokit_server- US:
ws://indie.echokit.dev/ws/ - Asia:
ws://hk.echokit.dev/ws/
- US:
Chat: press the K0 button once or multiple times util the screen shows "Listening ...". You can now speak and it will answer.
Record: long press the K0 until the screen shows "Recording ...". You can now speak and the audio will be recorded on the server.
Config: press RST. While it is restarting, press and hold K0 to enter the configuration mode. Then open the configuration UI to connect to the device via BT.