Crabnlp Speaker Diarization Using OpenAI Whisper

Crabnlp Transcription based on OpenAI Whisper and Custom Speaker Diarization

Please, star the project on github (see top-right corner) if you appreciate my contribution to the community!

What is it

Crabnlp is a transcription and speaker diarization pipeline based on OpenAI Whisper and custom speaker diarization.

Installation

Install the requirements

cd crabnlp
pip install -e crabnlp

Example using transcribe

from crabnlp.transcribe import transcribe

transcription_result = transcribe('path/to/file')

Known Limitations

• The system may not perform well in scenarios with overlapping speech or rapid speech.
• The system may not accurately identify speakers in scenarios with
• There might be some errors, please raise an issue if you encounter any.

Future Improvements

• Improvement the quality of sound recognition

Acknowledgements

This work is based on OpenAI's Whisper , Faster Whisper , Nvidia NeMo , and Facebook's Demucs