search

alumae / kaldi-gstreamer-server

Real-time full-duplex speech recognition server, based on the Kaldi toolkit and the GStreamer framwork.
BSD 2-Clause "Simplified" License
1.07k stars 341 forks source link

Invalid parameters supplied to OnlineLdaInput #147

Closed fancyerii closed 5 years ago

fancyerii commented 6 years ago

I trained the model by the recipe in the kaldi/egs/thchs30/s5 After training, I want to use kaldi-gstreamer-server to provide decoding service. my yaml is configed as: timeout-decoder : 10 decoder: model: test/models/thchs30/tri4b/final.mdl lda-mat: test/models/thchs30/tri4b/final.mat word-syms: test/models/thchs30/tri4b/words.txt fst: test/models/thchs30/tri4b/HCLG.fst silence-phones: "1" out-dir: tmp

When I run python kaldigstserver/client.py -r 32000 test/data/D21_848.wav The worker print error messages and exit abnormally.

ERROR ([5.4.232-532f3]:OnlineLdaInput():online-feat-input.cc:209) Invalid parameters supplied to OnlineLdaInput

[ Stack-Trace: ]

kaldi::MessageLogger::HandleMessage(kaldi::LogMessageEnvelope const&, char const) kaldi::MessageLogger::~MessageLogger() kaldi::OnlineLdaInput::OnlineLdaInput(kaldi::OnlineFeatInputItf, kaldi::Matrix const&, int, int)

clone

terminate called after throwing an instance of 'std::runtime_error' what():

alumae commented 6 years ago

Probably you are using pitch features, and the GMM-based decoder doesn't support pitch. Use nnet3-based model, then you can use pitch features.