Microsoft's AI tool can turn photos into realistic videos of people talking and singing – Engadget

Microsoft Research Asia has unveiled a new experimental AI tool called VASA-1 that can take a still image of a person — or the drawing of one — and an existing audio file to create a lifelike talking face out of them in real time. It has the ability to generate facial expressions and head motions for an existing still image and the appropriate lip movements to match a speech or a song. The researchers uploaded a ton of examples on the project page, and the results look good enough that they could fool people into thinking that they’re real.
While the lip and head motions in the examples could still look a bit robotic and out of sync upon closer inspection, it’s still clear that the technology could be misused to easily and quickly create deepfake videos of real people. The researchers themselves are aware of that potential and have decided not to release “an online demo, API, product, additional implementation details, or any related offerings” until they’re sure that their technology “will be used responsibly and in accordance with proper regulations.” They didn’t, however, say whether they’re planning to implement certain safeguards to prevent bad actors from using them for nefarious purposes, such as to create deepfake porn or misinformation campaigns.
The researchers believe their technology has a ton of benefits despite its potential for misuse. They said it can be used to enhance educational equity, as well as to improve accessibility for those with communication challenges, perhaps by giving them access to an avatar that can communicate for them. It can also provide companionship and therapeutic support for those who need it, they said, insinuating the VASA-1 could be used in programs that offer access to AI characters people can talk to.
According to the paper published with the announcement, VASA-1 was trained on the VoxCeleb2 Dataset, which contains “over 1 million utterances for 6,112 celebrities” that were extracted from YouTube videos. Even though the tool was trained on real faces, it also works on artistic photos like the Mona Lisa, which the researchers amusingly combined with an audio file of Anne Hathaway’s viral rendition of Lil Wayne’s Paparazzi. It’s so delightful, it’s worth a watch, even if you’re doubting what good a technology like this can do.
The US government is selling the Cheyenne Supercomputer at auction for a fraction of the list price right now.
A federal judge has sentenced Binance founder Changpeng Zhao to four months in prison. Prosecutors had recommended three years. Zhao pleaded guilty in November to violating the Bank Secrecy Act.
The newest Assassin’s Creed game will soon arrive on iPhone and iPad. Assassin’s Creed Mirage, the 2023 installment that takes you to ninth-century Baghdad, will be available on June 6 for the iPhone 15 Pro series and iPads with an M-series chip.
The National Highway Traffic Safety Administration just announced new safety standards for US cars. All new cars must have automatic braking systems installed by 2029.
The EU has officially opened a significant investigation into Meta for election disinformation. While the European Commission’s statement doesn’t explicitly mention Russia, Meta told Engadget the EU probe targets the country’s Doppelganger campaign.
The popular Arc Browser, which was previously a Mac exclusive, is now fully available for Windows users. It was in beta the past several months.
Apple has scheduled an event for May 7 that'll more than likely focus on new iPads. Here's what we expect the company to show off.
Microsoft has officially announced the next Xbox Games Showcase. In a blog post, the company said the summer version will be on Sunday, June 9, at 10AM PT / 1PM ET.
Apple is likely to introduce a new Pencil next week, but the second-gen model is back on sale for an all-time low of $79.
Tesla has begun laying off more people, starting with two senior executives, with plans for hundreds more. CEO Elon Musk says the company has to get “absolutely hard core about headcount.”
With a matte black finish, the Insta mini 99 looks more professional than most of its predecessors. While there are no hybrid digital camera features, it delivers far more versatility than pretty much any other instant camera.
Beats' latest earbuds have the longest battery life of any it has ever made. Plus, they're budget friendly at $80.
Beats made significant updates to audio quality and battery life on the Solo 4, but the design is in need of a refresh.
Simogo knows how to make a damn fine puzzle game.
Instagram is overhauling its recommendation algorithm for Reels to boost “original content” in a move that will have significant implications for aggregator accounts and others that primarily repost other users’ work.
The Federal Communications Commission has slapped the largest mobile carriers in the US with a collective fine worth $200 million for selling access to their customers' location information without consent.
Hybrid work isn't going away any time soon. These are the best gadgets to gift recent graduates who split their time working from home and working in an office.
The biggest news stories this morning: The DJI Mini 4K is a $299 drone for beginners, OpenAI will train its AI models on the Financial Times’ journalism, Gadgets that make great Mother’s Day gifts.
Yelp Assistant was trained on datasets from the website itself.
In order to figure out what the fastest charger on the market is across a range of power outputs, we tested 14 adapters across five different devices. Here are the results.
Subscribe to our two newsletters:
– A weekly roundup of our favorite tech deals
– A daily dose of the news you need
Please enter a valid email address
Please select a newsletter
By subscribing, you are agreeing to Engadget's Terms and Privacy Policy.


Leave a Reply

The Future Is A.I. !
To top