Route voice between Minecraft and external AI agents: send and receive audio via WebSocket so agents can listen and speak in-game through Simple Voice Chat.
Easy LLM Voice is a Fabric mod that connects Minecraft’s Simple Voice Chat to external AI agents over WebSocket. The mod forwards audio heard by players to an external program and plays audio received from that program inside the game, associating each audio stream with an in-game player so distance attenuation and audible range from Simple Voice Chat are preserved. A sample agent implementation is available: https://github.com/Koichiro-terao/Easy-LLM-Voice-Agent-in-Minecraft. Features include routing Simple Voice Chat audio to an external WebSocket server, playing incoming audio at the target player’s position, and managing multiple agent endpoints from an in-game GUI. Requirements: Minecraft, Java, Fabric Loader (or later), Fabric API, Simple Voice Chat, and an external WebSocket server program (for example, a Python-based agent). To install, place the Easy LLM Voice mod, Easy LLM mod, Fabric API, and Simple Voice Chat into the server-side mods folder; if you want to configure connection destinations from the GUI, also run the mod on the Minecraft client used for configuration. Connection model: Minecraft acts as a WebSocket client, so start a WebSocket server in your external program and register its host, port, and the associated player name via the mod GUI. The GUI accepts the Minecraft player name, hostname or IP, and port number for each endpoint. Usage outline: install required mods on the server, start your external program, join the server and press the V key to open the GUI (change the binding if it conflicts with Simple Voice Chat), add the connection by entering player name, host, and port, then select and activate the endpoint. Once connected, the external program can receive audio that the target player can hear and play audio from that player’s position. Note: this mod does not perform LLM processing, speech recognition, or speech synthesis itself; you must provide LLMs, ASR, TTS, Mineflayer, or other components on the external program side.
