Ilya Platonov
Ilya Platonov
Unfortunately this project is unable to keep up with all kaldi changes. So we probably should reference "old" kaldi build that supported by this server or find maintainer who can...
asr-server code is pretty old by now, I assume kaldi had new types of models. There could be some incompatibility here or you just do not load it correctly.
Few years ago, when I tested this on usual EC2 instance, it was 1 core can manage one concurrent speech recognition in realtime. So 10 cores for 10 concurrent speech...
As far as I know GPU used exclusively for training.
You should address this question to kaldi project developers. This server was developed a while ago and not really in sync with latest kaldi developments.
ASR server keep one model in memory and serves it. It work as HTTP based wrapper around kaldi decoder. You can use it as codebase for your solution, but you...