Ateme and Lingopal.ai developed a simpler way to localise content for global audiences, using NVIDIA L4 GPUs and enabling multilingual content delivery for broadcasters and D2C services.
Ateme and Lingopal.ai language translation platform have together developed a simpler way to localise content for global audiences, with support from NVIDIA.
This collaboration merges Ateme's video compression and delivery expertise with Lingopal's AI-driven translation capabilities, aiming to support Direct-to-Consumer (D2C) video services and broadcasters to make content available in multiple languages.
Used as a service, Lingopal.ai involves no code and can handle any media format. Lingopal receives live feeds, sending them back translated in real time. Its AI engine has been trained on millions of data points in order to compete with computer human translators. Each voice is cloned in real time to match the original speaker’s voice and his or her manner of using slang, idioms and so on.
Lingopal accepts feeds of all types. Integrated into workflows through APIs, the service can be connected to a live stream and customised for particular requirements.
Ateme’s delivery software streams the highest quality content at the lowest bitrate, and supports UHD, 4K and HDR. Their Kyrion and TITAN encoders/decoders can be used to optimise delivery platforms by repurposing available cloud-native resources for file transcoding.
Kyrion and TITAN aim to keep latency low on all platforms and maintain broadcast-level latency on streaming platforms, with reduced rebuffering. They use efficient compression that give users faster-than-real-time capabilities with simultaneous processing of multiple formats, and help save on bandwidth and storage requirements.
Using the hardware AI capabilities of NVIDIA’s L4 Tensor Core GPU, the Ateme-Lingopal solution can generate translations in over 120 languages. The two companies have aimed to design a system that preserves the tone, nuance and emotion of the original audio mix as well, in order to enhance the viewer experience.
The NVIDIA L4 GPU runs on NVIDIA Ada Lovelace architecture. Its energy-efficient acceleration for video and graphics, AI, visual computing and virtualisation makes it suitable for streaming live to millions of viewers. L4 delivers much higher AI video performance than CPU-based solutions supporting, for example, real-time insights to personalise content, and improved search relevance.
"This partnership marks a significant step toward breaking language barriers in global content distribution," said Mickaël Raulet, CTO at Ateme. "By integrating AI-driven translation into our video delivery workflows, we are enabling content creators to have a better chance of reaching wider audiences with relative ease and quality."
Lingopal.ai CEO Deven Orie said, "Our AI-powered translation, combined with Ateme's robust video delivery platform, allows broadcasters to localise content at scale. With little more than the click of a button, a live broadcast can be heard in any language worldwide, clearing the way to new revenue streams and opportunities for viewership growth."
This approach to multilingual content distribution gives broadcasters and streaming platforms a cost-effective, scalable, high-quality approach to engaging global audiences. www.ateme.com
