Enable neural vad

with the quickstart scripts how do you enable marble telephony?

Hi @ryein

Thanks for your interest in Riva

I will check with the team and provide inputs

Thanks

HI @ryein

I checked with the team,

Currently we do not provide Marblenet telephony as a part of our quickstarts

You can deploy it using the servicemaker container and the steps in the riva documentation : Pipeline Configuration — NVIDIA Riva

Thanks

Is there any other information about this?

In my case, I have the ASR configured for Spanish. I started it using the quickstart guide, but many times it transcribes words from noise. I think that VAD could help, but it is not clear to me how to configure it.

Do I have to regenerate the riva-build speech_recognition for Spanish? What happens to the already deployed model? Should it be deleted? What are the default parameters I should use?