YouTube Expands Speech Recognition and Translation AI Features
Google has augmented and widened access to some of YouTube’s audio AI-based features. The update includes extending auto-captioning to any YouTube channel and automatic caption translation to mobile devices and lays out plans for even more inclusion of the platform’s speech recognition and translation technology.
The most notable immediate change is that YouTube has ended the 1,000 subscriber minimum to enable live auto captions. The limit on auto captioning may have been a way to encourage the promotion of YouTube channels or out of concern for limited computing resources, but that no longer matters. The auto captions will also be available in more languages soon, upping the accessibility of non-English content on YouTube. The automatic translation feature for the captions will also become more available with a mobile version of the feature coming to Android and iOS devices instead of requiring a desktop to get the AI-powered translation to work.
“Starting today, creators can now enable live auto captions for any live streams in English to make streams more inclusive and accessible!” YouTube explained in a blog post about the new features. “In the coming months, we plan to expand live auto captions to all 13 supported automatic captioning languages.”
Search and Transcribe
Google has a lot of experiments ongoing with YouTube. This month, YouTube started testing connecting its suggestions for videos made by Google Assistant to its Android app. Premium subscribers can take part in the experiment and get recommendations within the video description of whatever they are currently watching. Some videos will have a Google Assistant card they can tap to see a Knowledge Graph panel with information about the video and suggested related videos. YouTube Premium subscribers can also test out downloading videos from web browsers and a picture-in-picture (PiP) feature that lets users keep watching YouTube videos from a miniature window.
The automatic captioning and transcription features will also play a role in some of YouTube’s other upcoming plans, including permitting subtitle and caption creation to trusted users and an experiment where users can type in the search bar at the top of a transcript to find keywords The most ambitious current project related to multiple audio tracks on a video to make it more accessible.
“We’re testing, with a small group of creators, the ability to add multiple audio tracks on their videos which will help provide multi-language audio for international audiences as well as descriptive audio for people who are blind or low-vision. We hope to roll this feature out more widely in the coming quarters,” YouTube explained. “Improving accessibility is a top priority for YouTube, and we hope these updates will help creators reach a wider audience.”