The Morning After: Microsoft’s VALL-E AI can replicate a voice from a three-second sample
Microsoft’s latest research in text-to-speech AI centers on a new AI model, VALL-E. While there are already multiple services that can create copies of your voice, they usually demand substantial input. Microsoft claims its model can simulate someone’s voice from just a three-second audio sample. The speech can match both the timbre and emotional tone of the speaker – even the acoustics of a room. It could one day be used for customized or high-end text-to-speech applications, but like deepfakes, there are risks of misuse.
Researchers trained VALL-E on 60,000 hours of English language speech from 7,000-plus speakers in Meta’s Libri-Light audio library. The results aren’t perfect: Some are tinny machine-like samples, while others are surprisingly realistic.
Microsoft isn’t making the code open source, possibly due to the inherent risks. In the paper, the company said: “Since VALL-E could synthesize speech that maintains speaker identity, it may carry potential risks in misuse of the model, such as spoofing voice identification or impersonating.”
We’ve all seen the 1992 movie , right? Right?!
– Mat Smith
The biggest stories you might have missed
But you’ll need a Prime subscription.
Amazon is expanding Prime to cover more of the web. The company says it’s making Buy with Prime “widely available” to eligible third-party sites in the US on January 31st. More shops can offer free shipping, streamlined checkout and simplified returns to Prime members. The theoretical advantages are clear: You get products with less hassle, while stores are more likely to turn visitors into paying customers. Amazon, meanwhile, is hoping to boost interest in Prime subscriptions. The catch, of course, is you have to pay for that pesky Prime sub.
Including a Titan seaplane and faster deep space travel.
NASA is handing out $175,000 initial study grants to 14 new projects potentially useful for missions in and beyond the solar system. TitanAir might be the most unusual one: a seaplane from Planet Enterprises’ Quinn Morley that could fly through the nitrogen-and-methane atmosphere of Saturn’s biggest moon, Titan, and sail its oceans.
MIT’s Mary Knapp has proposed a deep space observatory that would use a swarm of thousands of tiny satellites to detect low-frequency radio emissions from the early universe, and UCLA’s Artur Davoyan’s idea could speed up exploration at the outer edges of space. His design would propel spacecraft with a “pellet-beam” of microscopic particles traveling at very high speed (over 74 miles per second) using laser blasts.
Virgin Orbit said the rocket suffered an anomaly that prevented it from reaching its destination.
Meanwhile, the UK’s first efforts at space flight have ended in failure. Virgin Orbit’s historic Start Me Up mission launched from Spaceport Cornwall on January 9th as planned, but it failed to reach orbit. Apparently, the company tweeted, because of “an anomaly.” The mission carried payload satellites from seven commercial and government customers. They include a joint UK–US project called CIRCE (Coordinated Ionospheric Reconstruction CubeSat Experiment) and two CubeSats for the UK’s Ministry of Defense. As noted by Ars Technica, this failure could have a huge impact on the company, which is struggling to launch enough missions to break even.
Broadcom and Qualcomm may get the boot very soon.
Bloomberg sources claim Apple is not only prepping its first cellular modem (now slated for late 2024 or early 2025) but also a combination of Bluetooth and WiFi chips to replace the Broadcom chip currently handling those duties in iPhones. While the exact reasoning for the transition wasn’t mentioned, it’s no secret Apple started designing its own silicon across multiple products.
All products recommended by Engadget are selected by our editorial team, independent of our parent company. Some of our stories include affiliate links. If you buy something through one of these links, we may earn an affiliate commission. All prices are correct at the time of publishing.