There are currently many artificial intelligence (AI) tools on the market that can take users’ text and images and transform them into images and videos that match the initial prompt. A new patent reveals that audio may soon be an input option to bring your visions to real life.
As spotted by MSPowerUser, the US Patent and Trademark Office (USPTO) posted a 20-page document filed by Microsoft on April 5, 2023, and published on October 10, 2024, that details a new AI-supported system that converts live audio into images.
Also: Adobe’s free AI video generator is here – how to try it out
This system would take an audio live stream, such as that from a meeting or lecture, and convert it into a live text transcript. The transcript would then be summarized by a large language model (LLM) and fed into a text-to-image model, where an image would be generated and output on the screen, as seen in the image below.
<!–>
–>
Microsoft
This system would continue to do this during the audio stream, continuously generating live images. According to Microsoft, displaying images in real-time can help make communication more effective, with visual aids keeping people more engaged and making concepts easier to understand.
<!–>
“Displaying images related to verbally communicated information can enhance the effectiveness of communication by making it more engaging, memorable, and easier to understand,” said Microsoft.
If you’re wondering whether the feature will launch soon, the answer is most likely no. Filing a patent is a long journey between producing a product or feature, and many patents never make it into the production phase and remain an idea.
However, if Microsoft does decide to launch this feature, it would likely live in Microsoft Teams, its video conferencing meeting platform, and be accessible through its AI add-on, Copilot, such as Copilot Pro or Microsoft 365 Copilot for businesses.
