Hi, I have a brand new AGX Orin 64GB. I have followed the instructions here, but when I generate TTS within the docker container after getting Riva started successfully (no errors), I get the following intermittent error, seemingly randomly:
Error: Triton model failed during inference. Error message: Streaming timed out
When I retry, it works sometimes.
Any help would be appreciated.
Hi @guitargice, if you are using streaming TTS, I believe that is a known issue with Riva Release 2.15:
https://docs.nvidia.com/deeplearning/riva/user-guide/docs/release-notes.html#known-issues
- The first Riva TTS call after
riva_start.sh results in longer latency, and can throw a timeout error on some GPUs. Subsequent calls will exhibit normal latency.
Hi Dusty! Wow that was an insanely quick response. Thank you! Sorry for missing a known issue, appreciate the tip.
Oh yea no worries, it should be fixed in the next release. I did get XTTS working which I am using in the meantime.