Ecoute is a live transcription tool designed for real-time transcription of both microphone input and system audio output, aiding users in conversations. It targets individuals needing immediate text feedback from their audio environment.

How It Works

Ecoute leverages the Whisper ASR model for transcription. By default, it uses the 'tiny' model for low resource consumption and fast responses, primarily supporting English. An optional --api flag enables the use of OpenAI's Whisper API, offering significantly improved speed, accuracy, and multi-language support, albeit at a higher cost due to API usage.

Quick Start & Requirements

• Install: pip install -r requirements.txt
• Prerequisites: Python >=3.8.0, Windows OS, FFmpeg (installable via Chocolatey: choco install ffmpeg), OpenAI API key (for --api flag).
• Run: python main.py or python main.py --api for Whisper API.
• Docs: Demo available at https://github.com/user-attachments/assets/5616421f-838d-439f-8b15-0df7b8d33459.

Highlighted Details

• Real-time transcription of microphone and speaker audio.
• Optional Whisper API integration for enhanced speed, accuracy, and multi-language support.
• Default 'tiny' Whisper model for low resource usage.

Maintenance & Community

Contributions are welcome via issues and pull requests.

Licensing & Compatibility

Licensed under the MIT License, permitting commercial use and closed-source linking.

Limitations & Caveats

Currently limited to default system microphone and speaker devices. The non-API 'tiny' Whisper model has reduced accuracy for accents and non-English languages. Multi-language support is pending for the local model.