Ecoute is a live transcription tool that captures and displays real-time transcripts of both microphone input and speaker output simultaneously in a text interface.
The tool solves the problem of following conversations by providing immediate, dual-channel transcription. It works by capturing audio from the user's microphone and speakers, then processing both streams through speech-to-text conversion. Users can run it with a local model for basic English transcription or use the optional Whisper API flag for faster, more accurate transcription across multiple languages, though the API option incurs additional costs.
Ecoute suits developers and users who need to monitor or record conversations in real-time, particularly those working on Windows systems. The tool is designed to assist during conversations by providing live transcripts that can help users follow discussions or generate response suggestions. The README identifies two operational modes: a local model approach for basic use, and the Whisper API approach for enhanced speed and multilingual support. The choice between them depends on whether accuracy and language coverage justify the additional API costs.
The project shows active maintenance with multiple execution paths available to users, including a standard mode and an optimized variant. The README acknowledges known limitations in the tool's functionality and explicitly notes that the Whisper API approach is expected to become the default option in future releases, indicating ongoing development direction. The tool requires Python, FFmpeg, and optionally an OpenAI API key, with setup instructions provided for Windows environments.
