I keep getting a runtime error and the service keeps shutting down. I have described what I did below.
The README said to get my own kaldi model and must have a yaml file that describes the model. So I got a this model tedlium-nnet2, and put it inside /media/kaldi_models , which is mirrored in /opt/models in the docker container.
So my model english lives at /opt/models/english.
For the yaml file, I copied the 3rd one in your example from here and changed all the test/models/english to just english, and renamed it to nnet2.yaml.
I also copied the sample_full_post_processor.py file from alumae/kaldi-gstreamer-server/master/ because the yaml file required it.
But when I start the service using the command /opt/start.sh -y /opt/models/nnet2.yaml, I get this error Error opening input stream test/models/english/tedlium_nnet_ms_sp_online/ivector_extractor/final.mat. Since I refactored all the paths, why is it still looking under test/models?
See the worker log below.
2018-12-09 06:04:28 - INFO: decoder2: Created GStreamer elements
2018-12-09 06:04:28 - DEBUG: decoder2: Adding <__main__.GstAppSrc object at 0x7f9e215e6550 (GstAppSrc at 0x160a1b0)> to the pipeline
2018-12-09 06:04:28 - DEBUG: decoder2: Adding <__main__.GstDecodeBin object at 0x7f9e215e6370 (GstDecodeBin at 0x1620090)> to the pipeline
2018-12-09 06:04:28 - DEBUG: decoder2: Adding <__main__.GstAudioConvert object at 0x7f9e215e6320 (GstAudioConvert at 0x162bee0)> to the pipeline
2018-12-09 06:04:28 - DEBUG: decoder2: Adding <__main__.GstAudioResample object at 0x7f9e215e6500 (GstAudioResample at 0x14e8bd0)> to the pipeline
2018-12-09 06:04:28 - DEBUG: decoder2: Adding <__main__.GstTee object at 0x7f9e215e6460 (GstTee at 0x163c000)> to the pipeline
2018-12-09 06:04:28 - DEBUG: decoder2: Adding <__main__.GstQueue object at 0x7f9e215e64b0 (GstQueue at 0x1640210)> to the pipeline
2018-12-09 06:04:28 - DEBUG: decoder2: Adding <__main__.GstFileSink object at 0x7f9e215e65a0 (GstFileSink at 0x1644c00)> to the pipeline
2018-12-09 06:04:28 - DEBUG: decoder2: Adding <__main__.GstQueue object at 0x7f9e215e65f0 (GstQueue at 0x1640500)> to the pipeline
2018-12-09 06:04:28 - DEBUG: decoder2: Adding <__main__.Gstkaldinnet2onlinedecoder object at 0x7f9e215e6640 (Gstkaldinnet2onlinedecoder at 0x16660a0)> to the pipeline
2018-12-09 06:04:28 - DEBUG: decoder2: Adding <__main__.GstFakeSink object at 0x7f9e215e6690 (GstFakeSink at 0x1541200)> to the pipeline
2018-12-09 06:04:28 - INFO: decoder2: Linking GStreamer elements
ERROR ([5.4.176~1-be967]:Input():kaldi-io.cc:756) Error opening input stream test/models/english/tedlium_nnet_ms_sp_online/ivector_extractor/final.mat
[ Stack-Trace: ]
kaldi::MessageLogger::HandleMessage(kaldi::LogMessageEnvelope const&, char const*)
kaldi::MessageLogger::~MessageLogger()
kaldi::Input::Input(std::string const&, bool*)
void kaldi::ReadKaldiObject<kaldi::Matrix<float> >(std::string const&, kaldi::Matrix<float>*)
kaldi::OnlineIvectorExtractionInfo::Init(kaldi::OnlineIvectorExtractionConfig const&)
kaldi::OnlineNnet2FeaturePipelineInfo::OnlineNnet2FeaturePipelineInfo(kaldi::OnlineNnet2FeaturePipelineConfig const&)
gst_pad_query
gst_pad_query_caps
gst_element_get_compatible_pad
.
.
.
python() [0x4b988b]
PyEval_EvalFrameEx
PyEval_EvalFrameEx
PyEval_EvalCodeEx
python() [0x50160f]
PyRun_FileExFlags
PyRun_SimpleFileExFlags
Py_Main
__libc_start_main
python() [0x497b8b]
terminate called after throwing an instance of 'std::runtime_error'
what():
I keep getting a runtime error and the service keeps shutting down. I have described what I did below.
The README said to get my own kaldi model and must have a yaml file that describes the model. So I got a this model tedlium-nnet2, and put it inside
/media/kaldi_models
, which is mirrored in/opt/models
in the docker container.So my model
english
lives at/opt/models/english
.For the yaml file, I copied the 3rd one in your example from here and changed all the
test/models/english
to justenglish
, and renamed it tonnet2.yaml
.I also copied the
sample_full_post_processor.py
file fromalumae/kaldi-gstreamer-server/master/
because the yaml file required it.But when I start the service using the command
/opt/start.sh -y /opt/models/nnet2.yaml
, I get this error Error opening input stream test/models/english/tedlium_nnet_ms_sp_online/ivector_extractor/final.mat. Since I refactored all the paths, why is it still looking undertest/models
?See the worker log below.