A voice model running on CPU, inside a game, on a laptop.
Every line generated locally, leaving the GPU free for the game.
We built Space Jarv with @GradiumAI’s Phonon and Xybrid. Every character could have a voice, even offline.
Proud of what we built together 🚀
First game built on Phonon.
The @xybrid_ai team built Space Jarv, a 3D game where Phonon runs inside the game on a 16GB M1 Pro MacBook.
Phonon voices every line live on the player’s device.
If your game has talking characters, give them Phonon.
https://t.co/2YpIsrivv1
This is Phonon in action. We built a voice controlled rover that talks back.
Congratulations to the @GradiumAI team for the launch. 🚀 🌔
Excited to have partnered and brought Day-0 support on Xybrid.
Voice-controlled 3D games are here.
We partnered with @GradiumAI to give our game a voice: Phonon, their new 55M-parameter model, running next to an 8B agent and Whisper.
Inference optimized for CPU. Wi-Fi off.
Running on 16GB M1 Pro MacBook.
No cloud.
A new frontier for on-device speech.
We're open-sourcing Phonon, a suite of TTS models under 50m parameters, setting a new standard for naturalness and accuracy across languages, with the broadest inference support to date.
Phonon-31m is available under Apache 2.0.
https://t.co/d3qa9qmnzY🧵
About 15 voice commands and tool calls in one run: survey, atmosphere scans, drone camera, navigation, lift-off. A fully local agentic loop. At a live event, strangers asked for things we never scripted ("explore the area") and it still understood.
Alongside Phonon, Whisper turns your voice into text and LFM2.5-8B-A1B (@liquidai) runs the agent and calls the game's tools. Xybrid runs all three models alongside a full 3D Unity game on one 16GB M1 Pro MacBook.