How to test riva VAD model separately

Is there any way to test riva VAD model separately with the audio file? Actually I want to test Riva Marblenet Voice Activity Detection | NVIDIA NGC

with audio file

Hi @mehadi.hasan

Thanks for your interest in Riva

I will check with the team and provide the inputs

Thanks

Hi @mehadi.hasan

I have inputs from the team

We have an equivalent nemo vad model is here: VAD Marblenet | NVIDIA NGC
So this nemo model can be used to test VAD on it’s own

Thanks

Hi @rvinobha Thank you for your reply.

According to you the model

  1. Riva Marblenet Voice Activity Detection | NVIDIA NGC
  2. VAD Marblenet | NVIDIA NGC
    are equivalent.

Could you please check the dataset section of both model links? In the documentation, I’m seeing the Riva model is trained with 5 different datasets. On the other hand, the equivalent nemo model trained with only 2 datasets?

Do I miss something?