transcribe
Seamless voice input for the desktop. A local transcription router that converts your speech to text and pipes it directly into the agent.
Fast Local STT
By utilizing highly optimized local speech-to-text models like Whisper, hirn_transcribe allows you to talk to your desktop agent in real-time. It effortlessly captures complex coding terms, shell commands, and casual requests.
Absolute Privacy Guarantee
Unlike cloud dictation services, your voice never leaves your machine. Audio is processed completely locally and securely discarded. Only the transcribed text is sent to the LLM as a prompt, ensuring complete privacy over your ambient environment and conversations.
Always Ready
Accessible via a global hotkey, letting you instantly drop into a conversation with your agent without needing to switch windows.
Multi-lingual
Automatically detects your spoken language and seamlessly translates your speech to English code comments when needed.
VAD Optimized
Voice Activity Detection ensures the model only runs when you're actually speaking, saving massive amounts of battery and compute.