RecognitionConfig.CustomConfigurationEntry and other pipeline config documentation

AI & Data Science Deep Learning (Training & Inference) Riva
1

Riva Version: 1.8.0b0

I have noticed the continuous improvements in the documentation as the Riva team has been releasing newer versions (thank you very much!)

However, I still don’t understand what the following configurations do and/or mean:

  1. gRPC RecognitionConfig.CustomConfigurationEntry

    How is this parameter in the gRPC request used? Can I use it to specify whether I want a streaming or offline model for example? Or is this something that will be used in the future?

  2. riva-build speech-recognition’s padding_size parmeter

  3. VAD parameters. How does each parameter affect VAD?

1 Like
3

Hi @pineapple9011

  1. RecognitionConfig.CustomConfigurationEntry parameter is reserved for future use.
  2. riva-build speech-recognition 's padding_size parameter are values in seconds. It’s just how much forward and backward context is used. It’s padding if there’s insufficient data
  3. VAD parameters.
    Below section covers it’s affect:
    https://docs.nvidia.com/deeplearning/riva/user-guide/docs/service-asr.html?highlight=vad#beginning-end-of-utterance-detection

Thanks