An IIHT Company

NVIDIA® Riva stands as a GPU-accelerated software development kit (SDK) designed to create and deploy highly customizable, real-time, multilingual speech and translation AI pipelines. These pipelines encompass automatic speech recognition (ASR), neural machine translation (NMT), and text-to-speech (TTS) capabilities. Riva offers top-tier out-of-the-box precision in transcriptions and translations, along with engaging, professional-grade voices. Its deployment options include on-premises, across various cloud environments, embedded systems, and edge devices.

NVIDIA Riva empowers customers to tailor solutions for different languages, accents, domains, vocabularies, and contexts, ensuring optimal accuracy aligned with their specific use cases. It allows the customization of brand voice and intonation. Moreover, Riva ensures consistent experiences across a multitude of input streams, delivering superior inference performance compared to existing technologies.

To expedite the development of AI solutions based on speech, Riva provides pre-packaged AI workflows for tasks like audio transcription and intelligent virtual assistants. These workflows include pre-trained models and a wealth of resources, such as Helm Charts, Jupyter Notebooks, and documentation. Riva is an integral component of the NVIDIA AI Enterprise software platform.

Components within the NVIDIA Riva AMI comprise:

NVIDIA AI Enterprise Catalog access script
Access to the Riva container on NGC
Ubuntu Server 22.04
NVIDIA Container Toolkit
JupyterLab (accessible within the conda base environment)