Unable to Setup/Deploy ASR on AWS EKS

[bjoish]

Hi
So I am trying to set up/deploy ASR on AWS EKS using link 1…
there is this one thing I am not able to understand
So in step Defining the Client Service and Ingress Route, the first sub-step is to pull Riva-API-client Riva NGC. But if you look at the NGC container list, it’s not there. There are only two containers (Riva speech skills, Riva speech client).

In my NGC catalog:

The other thing is that when we come to step Defining and Launching the EKS Cluster, in sub-step 4, I am asked to change the directory to a folder called Riva-API-Nvidia and run a helm install. But where do I get this folder from is not clear?

Could anyone guide me to help resolve this?
Thank you

[SunilJB]

Hi @bjoish,

@bjoish

Riva-API-client Riva NGC

This is nothing but “Riva speech client” NGC container.

Thanks

[bjoish]

Hi @SunilJB,
Got it
I had another error popping up
when I trying to use this command

helm install --namespace riva riva --set ngcCredentials.password=echo -n $NGC_API_KEY | base64 -w0 --set modelRepoGenerator.modelDeployKey=echo -n tlt_encode | base64 -w0 --set riva.speechServices.asr=true --set riva.speechServices.tts=true --set riva.speechServices.nlp=true

It gives me an error
Error: failed to download "riva-api" (hint: running helm repo update may help)

I also tried updating the helm, but same error
Anything that I am missing or doing wrong…?

[SunilJB]

Hi @bjoish,

Below command should solve the issue, provided that you are in the extracted tar ‘riva-api’ folder containing values.yaml file:

helm install --namespace riva riva . --set ngcCredentials.password=`echo -n $NGC_API_KEY | base64 -w0` --set modelRepoGenerator.modelDeployKey=`echo -n tlt_encode | base64 -w0` --set riva.speechServices.asr=true --set riva.speechServices.tts=true --set riva.speechServices.nlp=true

Thanks