github·Robotics and devices
Voice-to-Intent Pipeline (Whisper + Jev)
Speech interface translating live microphone stream to typed hardware controls in under 80 milliseconds without waiting for an LLM answer.
Open GitHub ⚡ 72ms total latency
Map speech audio transcription to discrete device command opcode.
Jev sits between the incoming context and the next system action. Rather than generating lengthy, slow natural language that requires brittle regex parsing, Jev returns non-autoregressive, calibrated probabilities that downstream code can immediately execute.
01command_opcodePrimary action
02target_deviceConfidence score
03urgencyFallback route
Related Robotics and devices Builds
xpost
InspectVoice control over Fusion 360
I made this for fun. Wired a mic over Fusion 360: click, speak, Fusion runs the feature. @typesafeai ‘s Jev only decides if the utterance is a command.