Hi. I am trying to reproduce this tutorial (Tutorial page) and deploy a Conformer-CTC acoustic model with WFST decoders.
I downloaded the acoustic model from here RIVA Conformer ASR English - ASR set 6.0 | NVIDIA NGC
I immediately took the Conformer-CTC-L_spe128_en-US_6.0.riva model, so I started the tutorial from the build stage.
The build and deploy stages were successful. However, when executing the bash riva_start.sh i got this
Starting Riva Speech Services. This may take several minutes depending on the number of models deployed.
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Waiting for Riva server to load all models…retrying in 10 seconds
Health ready check failed.
Check Riva logs with: docker logs riva-speech
I executed the command docker logs riva-speech and found the following error in the logs:
E0227 08:29:32.956880 102 model_lifecycle.cc:596] failed to load 'asr_offline_kaldi_wfst_Conformer-CTC-L_spe128_en-US_6.0.riva-ctc-decoder-gpu-streaming-offline' version 1: Invalid argument: riva/cbe/asr/ctc-decoder/ctc-decoder.cc:73] Error occured during factory initialization: Kaldi decoder does not support subword decoding
Does anyone have any ideas what this could be about and what I should do in this case?
