soerson opened this issue 4 years ago · 0 comments
The text recognition is dramatically slow on the demo. It is far from real-time. Is there any way to make it closer to real-time transcribing?