Open-source local backend lets game NPCs talk to each other
A local speech-to-speech backend for game NPCs has been released as open source. , the , and speech generation all run on the computer, so no cloud service is needed. The focus is on NPCs talking to other NPCs, not only answering the player.
Each NPC can have its own , memory, and voice setup, and can later use earlier NPC-to-NPC conversations when speaking with the player. A background Game Manager AI can add hidden behavior notes to guide the story. In tests, for VR and a 7B model on a 4070 Ti reached about 400 to 600 ms TTFA, meaning the first audio started quickly enough to feel closer to a live conversation.
For multiple NPCs, a shared generation lock keeps the GPU from being overloaded by letting only one character generate speech at a time.
Key points
- The backend runs , the , and speech generation locally.
- NPCs can remember past conversations and use them later with the player.
- A Game Manager AI can quietly steer NPC behavior to shape the story.
- Tests reported about 400 to 600 ms TTFA before the first audio starts.
- A shared generation lock prevents several NPCs from overloading the GPU at the same time.