Get Started With NVIDIA Riva

NVIDIA Riva is a set of GPU-accelerated multilingual speech and translation microservices for building fully customizable, real-time conversational AI pipelines. Riva includes automatic speech recognition (ASR), text-to-speech (TTS), and neural machine translation (NMT) and is deployable in all clouds, in data centers, at the edge, or on embedded devices. With Riva, organizations can add speech and translation capabilities with large language models (LLMs) and retrieval-augmented generation (RAG) to transform chatbots into powerful multilingual assistants and avatars.

Riva provides deep learning-based automatic speech recognition (ASR), text-to-speech (TTS), and neural machine translation (NMT) models for AI practitioners and developers. ASR, TTS, and NMT are voice interfaces in speech AI-based applications, such as call center agent assists, digital assistants, video call transcriptions, and AI superchats driven by large language models (LLMs) and retrieval-augmented generation (RAG).

ASR converts speech to text and usually is the first step in a speech pipeline, so its transcription accuracy influences all downstream tasks. TTS generates human-like voices from text. NMT translates words from one language to another.

Riva is used across all industries—from telecommunications and finance to healthcare, retail, and automotive—wherever companies interact with customers.

Riva is part of the NVIDIA AI Enterprise software suite that includes business-standard enterprise support. Riva customers have priority access to new models, features, and supported releases with prioritized fixes.

The benefits include:

• World-class, real-time ASR in many languages, such as Arabic, Chinese (Mandarin), English (US/UK), French, German, Hindi, Italian, Japanese, Korean, Portuguese, Russian, and Spanish (LATAM/Spain), with full model customization to automate important processes with the best possible accuracy and unlock maximum business value.
• Expressive, professional, human-like TTS, out-of-the-box (OOTB) English (US/UK), German, Italian, Mandarin, and Spanish (LATAM/Spain) voices—female and male.
• High-quality OOTB bilingual and multilingual translation models and offline and streaming text-to-text, speech-to-text, and speech-to-speech support for up to 32 languages—Arabic, Bulgarian, Chinese, Croatian, Danish, Dutch, Estonian, Finnish, French, German, Greek, Hindi, Hungarian, Indonesian, Italian, Japanese, Korean, Latvian, Lithuanian, Norwegian, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Spanish, Swedish, Turkish, Ukrainian, and Vietnamese.
• Flexible deployment with consistent performance on premises, in all clouds, at the edge, and on the embedded devices.

NVIDIA Riva provides deep-learning-based ASR, NMT, and TTS skills for AI practitioners and developers. With Riva, you can:

• Voice your applications by using speech and translation AI skills in conversational applications across all industries, including AI superchats driven by large language models (LLMs) and retrieval-augmented generation (RAG).
• Create applications with engaging experiences by integrating world-class, OOTB ASR, TTS, and NMT skills and customizing models for the best possible transcription and translation accuracy and human-like expressivity for your use case.
• Offer highly accurate services to your customers by fine-tuning Riva models on your domain-specific data.

Riva is available as part of NVIDIA AI Enterprise. The full pricing and licensing details can be found here.

To learn more about purchasing Riva for production deployment, contact sales. Developers can also apply for a free 90-day trial of NVIDIA AI Enterprise to access the speech AI workflows, Riva containers, and pretrained models.

Reach out to your preferred NVIDIA partner to learn about options for purchasing NVIDIA AI Enterprise software. Independent software vendors (ISVs) should contact their regional NVIDIA sales representative, and partners can reach out to their NVIDIA business partner manager. If you have an existing speech AI project and would like to get started with testing and prototyping more quickly, you can request a free trial of Riva on NVIDIA LaunchPad.

NVIDIA LaunchPad is a universal proving ground that offers expansive testing of the latest NVIDIA enterprise hardware and software. This dynamic platform expedites short-term trials, facilitates long-term proofs of concept (POCs), and fuels the accelerated development of both managed services and standalone solutions. 

Users can initiate their AI journey with a prescriptive development environment tailored to their needs. Or they can explore a vast catalog of hands-on labs designed to offer immersive experiences across a spectrum of use cases, from AI-powered chatbots with NVIDIA Triton™ Inference Server to image classification with TensorFlow and more. Enterprises gain easy access to the latest accelerated hardware and software stacks deployed on privately hosted infrastructure.

Several labs use Riva in NVIDIA Launchpad. Organizations with an existing speech and translation AI project can apply to participate in the program free of charge. 

With LaunchPad, you don't need to have your own infrastructure or data to access the free trials. LaunchPad resources include NVIDIA-Certified Systems™ running complete NVIDIA AI software stacks, from GPU and DPU SDKs to leading AI frameworks and application frameworks. NVIDIA LaunchPad is available worldwide.

NVIDIA API catalog provides production-ready generative AI models and continually-optimized inference runtime, packaged up as microservices that can be easily deployed with standardized tools on any GPU-accelerated system.

NVIDIA AI Enterprise is an end-to-end, cloud-native software platform that accelerates data science pipelines and streamlines the development and deployment of production-grade AI applications, including generative AI, computer vision, speech AI, and many more. It includes best-in-class development tools, frameworks, pretrained models, and microservices for AI practitioners and reliable management capabilities for IT professionals to ensure performance, API stability, and security.