Microsoft has recently filed a patent application for an AI-powered technology that could transform how people experience live broadcasts and augmented reality content. According to the details, the system would allow AI agents to modify spoken words in near real time, replacing unwanted expressions, correcting speech errors and even translating languages while synchronising the speaker’s lip movements. The technology could make live content more accessible, personalised and easier to understand. However, the ability to alter what people hear and see also raises important questions about authenticity, accuracy and whether viewers will always know what was originally said. Here is everything you should know about Microsoft’s proposed AI system.
According to a report by Patentlyze, Microsoft has recently filed a patent application for a technology that would let near real-time AI enhance live video and augmented reality presentations. As per the details, the technology would use AI agents to identify words or phrases that match certain user settings. Once identified, the agents could then remove or replace them with different words or phrases.
The patent also suggests that the surrounding audio and video would be edited almost instantly to make the changes appear natural. Not only that, but the speaker’s facial movements and mouth would also be adjusted to match the new words. The system would reportedly work using pre-set rules or individual user profiles, giving viewers control over what they hear during a broadcast.
Also read: Karbonn Rebel Pro and Rebel launched in India: Check price, specs and more details
One of the proposed uses of the new technology is real-time language translation. The details suggest that it would allow viewers to watch a live event and hear a translated version in their preferred language, with the speaker’s mouth movements adjusted accordingly.
Furthermore, the patent filing also mentions removing filler words such as ‘um’ and ‘ah’ and correcting stutters and speech errors. This could also make the spoken content easier to follow.
Do note that the proposed changes go beyond basic corrections, as the system could also remove mumbled words, mispronounced words, misused words, misspoken words, offensive words, insensitive words, or inappropriate words. It may even replace certain words using suggestions from internet searches.
Also read: Portronics Air X air purifier is selling at a massive discount right now, check deal details
While the technology could make live content more accessible, it also raises concerns about authenticity. If viewers receive different versions of the same speech, they may not always know what the speaker actually said.
It could also lead to the removal of regional expressions and accents simply because they do not match a viewer’s preferences. However, the filing suggests that the original audio and video could be recorded and stored as a reference for accuracy.