Accelerating Conversational AI Research with New Cutting-Edge Neural Networks and Features from NeMo 1.0

Originally published at: https://developer.nvidia.com/blog/accelerating-conversational-ai-research-with-new-cutting-edge-neural-networks-and-features-from-nemo-1-0/

NVIDIA NeMo is a conversational AI toolkit built for researchers working on automatic speech recognition (ASR), natural language processing (NLP), and text-to-speech synthesis (TTS). The primary objective of NeMo is to help researchers from industry and academia to reuse prior work (code and pretrained models and make it easier to create new conversational AI models.…

it doesnt support nvidia Jetson embedded devices, does it?

Nowhere in the new User guide does it mention the platforms it does or does not run under…One would expect it to be under Prerequisites. Since there has not been any new Jetson downloads in many months don’t hold your breath!

@Andrey1984 and @Danieljyu2n, we currently don’t have official support of Nemo for Jetson embedded devices.

yes, it works somehow but it seems devs need to do support themselves to get it on arm

Good news and nice job Andrey…Did you need to do anything special that you want to share. If you say it “works” and I will take that with a grain of salt, then I will give it a go myself and report back my findings. It may help the Jetson team.

@Danieljyu2n
You may like to check NeMo installation - done; performance to be evaluated
However, for Jetson there is jetson-voice container at NGC that is poorly supported for recent Jetson versions though, rather not supported at all, but it is kind of based on not-supported at at all for Jetsons Jarvis that is a production version of NeMo that doesn’t supportt officially Jetson arch either.