A simple open source Speech-To-Text app. Powered by Comtegra GPU Cloud and Whisper.
whisper-large-v3: 5 minutes of audio → 35 seconds of processing
whisper-large-v3-turbo: 5 minutes of audio → 8 seconds of processing
Made with ❤️ by Comtegra S.A.
- Streamlit - for the app frontend
- Python - for the app backend
- Comtegra GPU Cloud (CGC) - for the best cloud compute
- Whisper-large-v3, Whisper-large-v3-turbo - for the speech to text models
- Improve how timestamps work
- Integrate additional/newer STT models
- Implement real-time transcription
- Option to transcribe only a part of the audio
- Add voice recording option
The sky's the limit! With CGC (Comtegra GPU Cloud), we're only limited by our imagination.
- Visit our usecase to see how to run the app and how it works
- Visit our documentation for more detailed information
- For questions or support, please contact: ai@comtegra.pl
This project is licensed under the Apache License 2.0. See the LICENSE file for details.
About Comtegra S.A.
Comtegra is an IT systems integrator based in Poland, specializing in various aspects of information technology, including data storage and management, information security, and network construction. Founded in 1999, Comtegra has established itself as a significant player in the Polish IT market, providing services such as backup solutions, cybersecurity, and virtualization technologies. The company emphasizes the integration of artificial intelligence within business operations to enhance data management and decision-making processes.