I'm on ubuntu 18.04, I work with Openvino 2020.1.023, Python* version 3.6.9 and Cmake* version 3.10.2. I tried to execute OfflineSpeech Recognition Demo with an other neuronal network than « lspeech_s5_ext » (which works either on CPU or VPU) then i tried with « wsj_dnn5b_smbr ». To do this, I modified the configuration file as attached.
Then, I execute this following command : ./offline_speech_recognition_app -wave="$wave" -c="$configuration_file"
Finally i got the following error message :
[ ERROR ] Add shift dimension mismatch (440 != 143)
Have you dot some documentation how to modify the configuration file for other featureTransform model than lspeech_s5_ext
You have incorrectly placed this query in the Intel® Compute Stick/Cards forum. To reach a more-appropriate audience, you need to move this query into the Intel® Neural Compute Sticks forum. Unfortunately, I do not have the capability to move this query for you; you will have to do this yourself.
We apologize for the inconvenience in our response. Could you please tell us if you were able to generate the slm.fst and labels.bin with the Kaldi model you used as stated in this documentation?
Also, could you provide us the link to the model you are using where we can find the HCLG.fst, transitions.txt and words.txt?