You might be talking in English, however to your colleague in Paris tuning into the Microsoft Groups assembly, you will sound such as you’re speaking in French.
Microsoft is at the moment testing a brand new Interpreter AI characteristic that clones your voice and converts it to a different language in real-time. The result’s a voice that sounds “similar to you in a distinct language,” based on the corporate. The translating program can be previewed early subsequent 12 months with as much as 9 languages, together with Italian, German, Japanese, Korean, Portuguese, French, English, Mandarin Chinese language, and Spanish. Solely accounts with a Microsoft 365 Copilot license will be capable to entry Interpreter, per The Washington Put up.
Microsoft’s AI enterprise is booming. CEO Satya Nadella stated on an earnings name final month that Microsoft’s AI division “is on monitor to surpass an annual income run price of $10 billion subsequent quarter” and grow to be “the quickest enterprise in our historical past to achieve this milestone.”
Microsoft Interpreter in Motion
In a single demo video, Interpreter interprets from Spanish to English in real-time in a Groups assembly, altering what the listener hears whereas sustaining the traits of the speaker’s voice.
In one other demo, Interpreter does the identical factor from English to Korean.
this is how the Microsoft Groups interpreter characteristic works to make it sound such as you’re talking in a international language on calls https://t.co/92al0jkG9u pic.twitter.com/B9zMLdFlBd
— Tom Warren (@tomwarren) November 19, 2024
Microsoft reassures customers that it’s going to not retailer their biometric info and can solely enable voice simulation with their consent.
The Execs and Cons of Voice Cloning
Voice cloning expertise is beneficial for extra than simply real-time interpretation. In July, AI startup ElevenLabs launched an app that contained the cloned voices of Judy Garland, James Dean, Burt Reynolds, and Sir Laurence Olivier. Customers may faucet into these voices to relate any e-book, doc, or file they uploaded.
There’s a draw back to the expertise, although: it makes scams all of the extra private. One AI cloning scheme copies somebody’s voice from simply three seconds of audio, like a video posted to social media. After cloning the voice, the fraudsters cold-call the sufferer’s family and friends to acquire cash.
Associated: Rising AI Menace Sounds Like Your Beloved One on the Cellphone — However It is Not Actually Them
Some AI firms have held again from releasing subtle voice cloning expertise as a result of it may very well be used for the incorrect functions. In April, ChatGPT-maker OpenAI introduced a Voice Engine AI generator that it stated may realistically mimic somebody’s voice from 15 seconds of audio — however determined to not broadly launch it due to “the potential for artificial voice misuse.”
