A cross-platform long-running microphone recorder with real-time transcription. It uses Qwen3-ASR by default. VAD keeps only speech segments and transcribes speech-only chunks. - View it on GitHub