This module uses OpenAI's gpt-live-transcribe streaming speech-recognition model: press Start Transcription, allow the browser to use your microphone, and just start talking — the transcript streams onto the panel as you speak, with very low latency. It is ideal for live meeting notes, lecture capture, classroom notes, interview shorthand, and on-site captions. Recognition handles multilingual and mixed-language speech, and sentence structure and punctuation are tidied automatically; after stopping, copy the full transcript for further editing. Press Stop Transcription when you are done. This module supports microphone input only — no text input or file upload; to transcribe an existing recording, use "Speech Recognition gpt_transcribe" instead. Please use a recent Chrome or Edge browser with microphone support. Output content formats: Text