An industrial-grade speech recognition toolkit boasting 170x realtime processing and support for over 50 languages with features like speaker diarization and emotion detection.
Discovered on GitHub via GitHub:modelscope