As a proof of concept, I'm making a C++ backend of ESPnet during my winter vacation. It can read ESPnet's dump features (json/scp) and train neural networks. Here is tiny AN4 example. Have fun.
https://github.com/ShigekiKarita/thxx/blob/master/example/speech_recognition/main.cpp
I found that current C++ backend is little bit faster than Python in training. (base transformer with minibatch 16)
30% speed up!
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.
This issue is closed. Please re-open if needed.
Most helpful comment
I found that current C++ backend is little bit faster than Python in training. (base transformer with minibatch 16)
30% speed up!