Speech sample application failing on SoftMax layer; this layer was removed previously using model optimizer.
git clone from https://github.com/openvinotoolkit/openvino
git rev-parse origin/master
fc3f9af923a1bd8ee2c232f227144f941f1f4015
GNA_AUTO run:
..\openVino\bin\intel64\Release>speech_sample -d GNA_AUTO -bs 2 -i dev93_10.ark -m wsj_dnn5b.xml -o scores.ark -r dev93_scores_10.ark
[ INFO ] InferenceEngine:
API version ............ 2.1
Build .................. custom_master_fc3f9af923a1bd8ee2c232f227144f941f1f4015
Description ....... API
[ INFO ] Parsing input parameters
[ INFO ] No extensions provided
[ INFO ] Loading Inference Engine
[ INFO ] Device info:
GNA
GNAPlugin version ......... 2.1
Build ........... custom_master_fc3f9af923a1bd8ee2c232f227144f941f1f4015_with_GNA_LIB_VER==2
[ INFO ] Loading network files
[ INFO ] Batch size is 2
[ INFO ] Using scale factor of 2175.43 calculated from first utterance.
[ INFO ] Loading model to the device
[ ERROR ] [GNAPlugin] in function void __cdecl GNAPluginNS::GNAPlugin::LoadNetwork(class InferenceEngine::ICNNNetwork &): The plugin does not support layer: softmax:SoftMax
Model optimizer removing softmax:
python3 mo.py --framework kaldi --input_model wsj_dnn5b.nnet --counts wsj_dnn5b.counts --remove_output_softmax
[ SUCCESS ] Generated IR version 10 model.
[ SUCCESS ] XML file: ..\openVino\model-optimizer\.\wsj_dnn5b.xml
[ SUCCESS ] BIN file: ..\openVino\model-optimizer\.\wsj_dnn5b.bin
[ SUCCESS ] Total execution time: 44.45 seconds.
I also tried by downloading the samples directly from 101.org. same behavior.
Link Copied
Hi Owais,
I’ve validated the Automatic Speech Recognition C++ Sample as per the instructions in the sample page, on Intel® Distribution of OpenVINO™ Toolkit.
https://docs.openvinotoolkit.org/2020.4/openvino_inference_engine_samples_speech_sample_README.html
I suggest you try running the sample from a different branch, or on OpenVINO 2020.4, which is our latest offering with improved performance.
Regards,
Munesh
Hi Owais,
I’ve validated the Automatic Speech Recognition C++ Sample as per the instructions in the sample page, on Intel® Distribution of OpenVINO™ Toolkit.
https://docs.openvinotoolkit.org/2020.4/openvino_inference_engine_samples_speech_sample_README.html
I suggest you try running the sample from a different branch, or on OpenVINO 2020.4, which is our latest offering with improved performance.
Regards,
Munesh
I pulled from git. did that work for you? or is the git solution non functional? What branch do you recommend?
Hi Owais,
We usually recommend validated branches, so as to avoid bugs, exception messages, etc. You can try using releases/2020/4, which is a validated branch.
Regards,
Munesh
Hi Owais,
This thread will no longer be monitored since we have provided a solution. If you need any additional information from Intel, please submit a new question.
Regards,
Munesh
For more complete information about compiler optimizations, see our Optimization Notice.