There are at the moment many synthetic intelligence (AI) instruments available on the market that may take customers’ textual content and pictures and remodel them into pictures and movies that match the preliminary immediate. A brand new patent reveals that audio could quickly be an enter choice to deliver your visions to actual life.
As noticed by MSPowerUser, the US Patent and Trademark Workplace (USPTO) posted a 20-page doc filed by Microsoft on April 5, 2023, and printed on October 10, 2024, that particulars a brand new AI-supported system that converts stay audio into pictures.
This technique would take an audio stay stream, comparable to that from a gathering or lecture, and convert it right into a stay textual content transcript. The transcript would then be summarized by a big language mannequin (LLM) and fed right into a text-to-image mannequin, the place a picture can be generated and output on the display, as seen within the picture beneath.
This technique would proceed to do that through the audio stream, constantly producing stay pictures. In response to Microsoft, displaying pictures in real-time might help make communication more practical, with visible aids protecting individuals extra engaged and making ideas simpler to know.
“Displaying pictures associated to verbally communicated info can improve the effectiveness of communication by making it extra partaking, memorable, and simpler to know,” stated Microsoft.
Should you’re questioning whether or not the characteristic will launch quickly, the reply is almost certainly no. Submitting a patent is an extended journey between producing a product or characteristic, and plenty of patents by no means make it into the manufacturing section and stay an concept.
Nevertheless, if Microsoft does determine to launch this characteristic, it might seemingly stay in Microsoft Groups, its video conferencing assembly platform, and be accessible by means of its AI add-on, Copilot, comparable to Copilot Professional or Microsoft 365 Copilot for companies.