🎤

Voice Cockpit

A fully interactive car cockpit controlled by voice. Speak naturally to adjust climate, control media, open windows, or set navigation. The full pipeline runs locally: LFM2.5-Audio-1.5B handles speech-to-text and text-to-speech, while LFM2-1.2B-Tool routes spoken intent to 26 cockpit functions. No cloud round-trip.

Full pipeline, no cloud — STT, intent parsing, tool execution, and TTS all run locally. No network dependency, no data leaves the device

Sub-200ms end-to-end — From spoken command to audible response in under 200ms. Cloud pipelines take 500ms-1s

26 functions, open vocabulary — Understands natural language across climate, media, navigation, and window controls. No rigid command grammar

The Problem

Cloud voice assistants add 500ms+ network round-trips for every interaction. On-device alternatives sacrifice accuracy. Automotive and embedded systems need both low latency and high accuracy without cloud dependency.

How LFM Compares

Cloud pipelines (Google STT + Dialogflow + Google TTS) deliver accuracy but add 500-1000ms latency and require connectivity. On-device keyword spotters are fast but limited to fixed vocabularies. LFM runs the full voice pipeline locally in under 200ms.

What LFM Unlocks

Complete voice AI on a single GPU: speech-to-text, semantic tool calling across 26 functions, and natural text-to-speech. Works offline, responds in real time, and understands open-ended commands.

This demo is fine-tuned on sample data. Results improve with your data.

Build Your Model on Workbench →Talk to Our Team