I am trying to run the basic riva speach example on the orin nx …
It looks like the riva-speech container is not firering up while running riva_start.sh
This is what i have done:
- Reinstalled the jetson to the newest Jetpack version…
- changed docker default runtime …
{
“default-runtime”: “nvidia”,
“runtimes”: {
“nvidia”: {
“args”: [ ],
“path”: “nvidia-container-runtime”
}
}
}
- created an API Key and logging in into docker like this: Step #2: Authenticating local Docker with NGC - NVIDIA Docs Which key is the correct on to use?! Personal or Leagacy?! (I messed up quite a bit on this step. Could the probleme come from here?)
- installed ngc like this: https://org.ngc.nvidia.com/setup/installers/cli Arm64Linux
echo “export PATH="$PATH:$(pwd)/ngc-cli"” >> ~/.bash && source ~/.bash
instead of
echo “export PATH="$PATH:$(pwd)/ngc-cli"” >> ~/.bash_profile && source ~/.bash_profile
As recomended on linux i used
- downloaded riva_quickstart_arm64:2.19.0
- Initialized stuff with riva_init.sh. After that i got the model_repository/models folder with models inside
- Try to start service eith riva_start.sh
No container is starting (docker ps is not showing a running container).
No logs providet via docker logs riva-speech.
Jetson info:
Thx 4 help.