NVIDIA RIVA: What's New
NVIDIA® Riva, part of the NVIDIA AI platform, is available with fully optimized pipelines for deploying real-time speech AI applications such as transcription and virtual assistants.
- Support for 2 new languages: Hindi and French
- Streaming accuracy improvement with Conformer CTC
- Word/profanity filter
- Lowcode AI custom voice creation with 30-minute voice input and controlled voice pitch, volume, and pause output
- Compact and simple model with multiple synthetic voices selection at inference time
- ASR in 7 languages and 2 out-of-the-box female and male English TTS voices
- Real-time performance with below 100 ms latency
- Runs on NVIDIA Jetson AGX Xavier, Jetson Xavier NX, Jetson AGX Orin and Jetson Orin NX
- Support for 5 new ASR languages: Arabic, Korean, Portuguese, Japanese and Italian
- Offline speaker diarization i.e. identifying speakers in the transcripts (up to 8 speakers)
- Speech hints API for improved accuracy on content such as addresses and credit cards
- TTS voice enhancement by controlling emphasis through SSML
Download Now Documentation Forum
Quick Start Guide
Get step-by-step instructions for deploying pretrained models as services on a local workstation and how to interact with them through a client.
Learn about Riva’s architecture, key features, and components for building speech AI services.
Riva Developer Starter Kits
Everything you need to start developing your speech AI with NVIDIA Riva, including tutorials, notebooks, and documentation.
Automatic Speech Recognition
- Speech Recognition Documentation
- Quick Start Guide
- Tutorial: Use Riva ASR Out-Of-The-Box
- Tutorial: Riva ASR Customization
- Sample App: Riva Contact
- Sample App: Virtual Assistant with Rasa
- TTS Documentation
- Quick Start Guide
- Documentation: TTS Service Sample
- Documentation: TTS SSML Sample
- Sample App: Virtual Assistant with RASA
NVIDIA’s platforms and application frameworks enable developers to build a wide array of AI applications. Consider potential algorithmic bias when choosing or creating the models being deployed. Work with the model’s developer to ensure that it meets the requirements for the relevant industry and use case; that the necessary instruction and documentation are provided to understand error rates, confidence intervals, and results; and that the model is being used under the conditions and in the manner intended.