NVIDIA at INTERSPEECH 2021

NVIDIA offers pre-trained models for speech recognition, language understanding, and speech synthesis through the NGC catalog. These models are highly accurate and have been trained on a variety of open and proprietary datasets for thousands of hours using GPUs. The NGC models are seamlessly integrated with SDKs such as NVIDIA NeMo for building, training, and fine-tuning conversational AI models

Explore NVIDIA NeMo, an open-source toolkit for researchers developing new state-of-the-art conversational AI models. It provides a collection of modules and models for automatic speech recognition, natural language processing, and text-to-speech. NeMo modules and models are highly interoperable with popular PyTorch and PyTorch Lightning frameworks, giving researchers exceptional flexibility.

NVIDIA offers Riva, a GPU- accelerated SDK to help enterprises develop multimodal conversational AI applications. It includes highly accurate pre-trained models in NGC, tools for fine-tuning these models on custom datasets, and optimized real-time speech and language skills for tasks like transcription and natural language-understanding.