1 comment

[ 3.1 ms ] story [ 13.9 ms ] thread
This one uses a simple, command based speech recognition system running on the browser - thanks to webassembly! The model is trained using Kaldi and PyTorch.