Advanced Television

Gcore launches AI Automated Speech Recognition

March 19, 2024

Gcore, the edge AI, cloud, network, and security solutions provider, has announced the public availability of its Gcore Artificial Intelligence Automated Speech Recognition (AI ASR). It seamlessly integrates into Gcore workflows enabling broadcasters, VoD, live streaming, and enterprise content owners to reach new global audiences, enhancing the accessibility of content for those speaking different languages or with hearing impairments.

Existing automated speech recognition (ASR) services can be slow and place a significant resource burden on content creators and owners. For broadcasters, enterprises or content owners with live news, sports events or investor relations information that must reach customers quickly, speed is essential. Traditional subtitle generation can take hours or even days if multiple languages are involved, and often results in inaccuracies.

Unlike other ASR services, Gcore AI ASR is a managed cloud service, supporting 100+ languages, that allows customers to focus on fast subtitle generation for their content, without the need for selecting and fine-tuning AI models. The Gcore team assesses newly released and updated ASR models, ensuring the best option is available through the pre-configured service. The managed service team supports customers with model selection and fine-tuning them to meet specific needs.

Gcore AI ASR generates subtitles for a one-hour video in under ten minutes, with accuracy levels matching or exceeding those of humans and typically achieving a 4 – 5 per cent word error rate. Open-source ASR models for specific languages or subject domains can be selected to enhance accuracy based on the content to be subtitled. This customisation is particularly useful for industry-specific terminology, or content featuring multiple spoken languages.

Alexey Petrovskikh, Head of Streaming Platform at Gcore, commented: “Subtitles are critical to reaching global audiences with content. Gcore’s AI speech recognition service – AI ASR – gives broadcasters, content owners and enterprises a cost-effective and accurate way to reach global audiences with fresh, accessible content. It is another step in our commitment to the continuous innovation of our solutions and edge infrastructure.”

Categories: Articles, Services

Tags: ,