Nvidia Partners with Rylo to Develop Real-Time AI Sign Language Translation
Nvidia is partnering with Rylo to develop real-time AI sign language translation. Concurrently, Rylo launched Rylo Desktop, a Mac app providing live meeting captions without external bots, following an $85 million funding round.
Nvidia is partnering with Rylo, formerly known as Nagish, to develop real-time sign language translation powered by artificial intelligence. Alongside this collaboration, the company is launching a new desktop application designed to make work meetings accessible for deaf and hard-of-hearing employees by generating real-time captions directly on the computer without requiring an external bot to join the call.
Advancing Sign Language AI with Nvidia
The collaboration with Nvidia focuses on accelerating the training of AI models capable of decoding and translating sign language video. The development relies on Nvidia's Cosmos Tokenizer technology, which enables efficient video representation and integrates into advanced model development workflows for real-time understanding, reconstruction, and translation. Model training is executed using Nvidia's computing infrastructure and software tools, including the NeMo Framework and Megatron Bridge.
This initiative also builds upon Rylo's acquisition of sign.mt less than a year ago. Dr. Amit Moryossef, founder of sign.mt, currently leads research at Rylo. According to Moryossef, developing models that reliably capture sign language nuances and convert them into text or speech in real time demands significant computing resources. He noted that Nvidia's infrastructure allows the company to train models at scale, with greater precision and at a much faster pace than previously possible.
Rylo Desktop: Captions Without External Bots
Simultaneously, Rylo is releasing Rylo Desktop, a free Mac application aimed at easing participation in work meetings for deaf and hard-of-hearing employees. Operating as an independent overlay above video conferencing platforms like Zoom, Microsoft Teams, and Google Meet, the app displays real-time subtitles and transcriptions.
A key feature of the application is that it functions directly from the computer, eliminating the need to invite an external bot to the meeting. Users can receive captions throughout the conversation without adding an automated participant to the attendee list. Tomer Aharoni, co-founder and CEO of Rylo, explained that the company aims to make communication accessible everywhere—in face-to-face conversations, on the phone, and in the workplace—through transcription, text-to-speech, and sign language translation.
"We strive to make communication accessible wherever it happens, whether in-person, on the phone, or in the workplace." — Tomer Aharoni, CEO of Rylo.
The application offers three display options: a scrollable chat-style transcription window, a large subtitle bar, or a minimalist Notch-style overlay that can be activated with a keystroke. Additionally, each speaker is identified by a permanent color to facilitate tracking the conversation. The app also includes a Type-to-Speak tool, allowing users to type text in real time and have it read aloud to other participants. Initially, the product is being offered in a private beta free of charge for Mac users in the United States.
Funding and Growth
These launches follow a recent $85 million funding round, bringing Rylo's total capital raised to $101 million. Specializing in AI-powered transcription and translation for the deaf and hard-of-hearing community, Rylo provides services for face-to-face and phone conversations, along with enterprise communication solutions and specialized cellular services. The company states it has already achieved profitability and estimates that its current growth trajectory will enable it to attain unicorn status—valued at over $1 billion—within roughly a year and a half. With offices in New York and Tel Aviv, Rylo holds two FCC certifications from the United States Federal Communications Commission.
