LFM2.5-Audio WebGPU

LFM2.5-Audio Demo

Speech recognition, text-to-speech, and conversational audio running entirely in your browser.

• ASR — Record or upload audio to transcribe to text
• TTS — Type text to generate natural speech
• Interleaved — Have a spoken conversation with the model (multi-turn with stateful KV cache)

Requires WebGPU-enabled browser (Chrome/Edge 113+). Enable at chrome://flags if needed. Use "New Chat" to reset conversation.