search

SlangLab-NU / VoiceCollector

Apache License 2.0
1 stars 1 forks source link

Model loading prior to starting the app #13

Open aanchan opened 1 year ago

aanchan commented 1 year ago

WWW As a developer running the app, the speech recognition model now is loaded in multiple places. It would be good to load the model once into memory prior to the start of the application before inference is run and not load it every time inference needs to be run. This makes things really slow. AC A pull-request with code changes where the model is loaded just once and used in multiple places wherever inference needs to be run.

aanchan commented 11 months ago

I am deleting some of the speech recognition code to make it all a bit lighter