| |
|
|
| |
|
|
Google Works Out a Fascinating, Slightly Scary Way For AI To Isolate Voices In a Crowd
https://arstechnica.com/gadgets/2018/04/google-works-out-a-fascinating-slightly-scary-way-for-ai-to-isolate-voices-in-a-crowd/
An anonymous reader quotes a report from Ars Technica:
Google researchers have developed a deep-learning system designed to help computers better identify and isolate individual voices within a noisy environment. As noted in a post on the company's Google Research Blog this week, a team within the tech giant attempted to replicate the cocktail party effect, or the human brain's ability to focus on one source of audio while filtering out others -- just as you would while talking to a friend at a party. Google's method uses an audio-visual model, so it is primarily focused on isolating voices in videos. The company posted a number of YouTube videos showing the tech in action.
The company says this tech works on videos with a single audio track and can isolate voices in a video algorithmically, depending on who's talking, or by having a user manually select the face of the person whose voice they want to hear. Google says the visual component here is key, as the tech watches for when a person's mouth is moving to better identify which voices to focus on at a given point and to create more accurate individual speech tracks for the length of a video. According to the blog post, the researchers developed this model by gathering 100,000 videos of "lectures and talks" on YouTube, extracting nearly 2,000 hours worth of segments from those videos featuring unobstructed speech, then mixing that audio to create a "synthetic cocktail party" with artificial background noise added. Google then trained the tech to split that mixed audio by reading the "face thumbnails" of people speaking in each video frame and a spectrogram of that video's soundtrack. The system is able to sort out which audio source belongs to which face at a given time and create separate speech tracks for each speaker. Whew. |
|
|
|
| |
|
| There are no conversations. |
| |
|
|
|
|
| M. H. Abrams |
We worked on solving the problem of voice communications in a noisy military environment. We established military codes that are highly audible and invented selection tests for personnel who had a superior ability to recognize sound in a noisy background. |
| James Fallows |
The demise of Google Reader, if logical, is a reminder of how far we've come from the cuddly old 'I'm Feeling Lucky' Google days, in which there was a foreseeably-astonishing delight in the way Google's evolving design tricks anticipated what users would like. |
| Martin Campbell |
I like pre-production and post the best. I don't like shooting at all. I find it grueling and tough, but I love post and the whole process of seeing the film finally come together. You start ironing out all the rough spots, and the really bad bits you just throw away. So from day one of post to the last day, you see nothing but improvements. |
| Josh Hamilton |
This is how I feel about horror films: there's enough scary things that happen in day-to-day life. Sometimes just going and getting the mail is scary, when you open your bills. And so, sometimes I feel like scary movies are just tapping into those anxieties and magnifying them. |
| Robert Darnton |
In 2002, Google began an ambitious project to digitize every book in the world. It was intended as a search project: type in a query, and Google would show you snippets. They asked university libraries for books, which they would scan for free. At Harvard we didn't permit them to take works under copyright, but other libraries gave them everything. |
| Marc Andreesen |
Google is working on self-driving cars, and they seem to work. People are so bad at driving cars that computers don't have to be that good to be much better. |
| Jim Garrison |
Until as recently as November of 1966, I had complete faith in the Warren Report. Of course, my faith in the Report was grounded in ignorance, since I had never read it. |
| Steve Israel |
The Stem Cell Research Enhancement Act would expand research on embryonic stem cells by increasing the number of lines stem cells that would be eligible for federally funded research. |
| Mike Ferguson |
America's doctors, nurses and medical researchers are the best in the world, but our health care system is broken. |
| Niger Innis |
When you take out individual initiative, individual responsibility, and the hope that every individual is born with, to better their lives, to climb the economic ladder, to pursue happiness, that is, in fact, a neoslavery. |
|
|
|
|
oogle's Voice-Generating AI Is Now Indistinguishable From Humans
Anonymous Coward 6 hours ago 75
An anonymous reader quotes a report from Quartz: A research paper published by Google this month -- which has not been peer reviewed -- details a text-to-speech system called Tacotron 2, which claims near-human accuracy at imitating audio of a person speaking from text. The system is Google's second official generation of the technology, which consists of two deep neural networks. The first network translates the text into a spectrogram (pdf), a visual way to represent audio frequencies over time. That spectrogram is then fed into WaveNet, a system from Alphabet's AI research lab DeepMind, which reads the chart and generates the corresponding audio elements accordingly. The Google researchers ...
|
| |
|
|
|
Is Google paying academics to only research topics it agrees with?
Google is being accused of using its funding power to push forward hundreds of research papers that support its agenda and business practices, particularly those that face criticism from regulators.
The Wall Street Journal (WSJ) has seen thousands of emails detailing financial relationships between Google and at least a dozen university professors from top-ranking universities in the world.
...
|
| |
|
|
|
Scientists Train AI To Learn People's Voices, Then Generate Their Faces
An neural network named "Speech2Face" was trained by scientists on millions of educational videos from the internet that showed over 100,000 different people talking. From this dataset, Speech2Face learned associations between vocal cues and certain physical features in a human face, researchers wrote in a new study. The AI then used an audio clip to model a photorealistic face matching the voice, and the results are surprisingly close to the actual faces of the people whose voices it listened to. The faces generated by Speech2Face didn't precisely match the people behind the voices. But the images did usually capture the correct age ranges, ethnicities and genders of the individuals, according to the study.
...
|
| |
|
|
|
Google Says Almost All CPUs Since 1995 Vulnerable To 'Meltdown' And 'Spectre' Flaws
Google has just published details on two vulnerabilities named Meltdown and Spectre that in the company's assessment affect "every processor [released] since 1995." Google says the two bugs can be exploited to "to steal data which is currently processed on the computer," which includes "your passwords stored in a password manager or browser, your personal photos, emails, instant messages and even business-critical documents." Furthermore, Google says that tests on virtual machines used in cloud computing environments extracted data from other customers using the same server. The bugs were discovered by Jann Horn, a security researcher with Google Project Zero, Google's elite security team. These are the ...
|
| |
|
|
|
Contractors Lose Jobs After Hacking CIA's In-House Vending Machines
An anonymous reader quotes a report from TechRepublic:
Today's vending machines are likely to be bolted to the floor or each other and are much more sophisticated -- possibly containing machine intelligence, and belonging to the Internet of Things (IoT). Hacking this kind of vending machine obviously requires a more refined approach. The type security professionals working for the U.S. Central Intelligence Agency (CIA) might conjure up, according to journalists Jason Leopold and David Mack, who first broke the story A Bunch Of CIA Contractors Got Fired For Stealing Snacks From Vending Machines. In their BuzzFeed post, the...
|
| |
|
|
|
The CCleaner Malware Fiasco Targeted at Least 20 Specific Tech Firms
Hundreds of thousands of computers getting penetrated by a corrupted version of an ultra-common piece of security software was never going to end well. But now it's becoming clear exactly how bad the results of the recent CCleaner malware outbreak may be. Researchers now believe that the hackers behind it were bent not only on mass infections, but on targeted espionage that tried to gain access to the networks of at least 20 tech firms. Earlier this week, security firms Morphisec and Cisco revealed that CCleaner, a piece of security software distributed by Czech company Avast, had been hijacked by hackers and loaded with a backdoor that evaded the company's security checks. It wound up installed on more than 700,000 co...
|
| |
|
|
|
Researchers Created ‘Quantum Artificial Life’ For the First Time
“Our research brought these amazingly sophisticated events called life to the realm of the atomic and microscopic world …and it worked.”
For the first time, an international team of researchers has used a quantum computer to create artificial life—a simulation of living organisms that scientists can use to understand life at the level of whole populations all the way down to cellular interactions.
...
|
| |
|
|
|
Prisons Across the US Are Quietly Building Databases of Incarcerated People's Voice Prints (theintercept.com)
In New York and other states across the country, authorities are acquiring technology to extract and digitize the voices of incarcerated people into unique biometric signatures, known as voice prints. From a report:
Prison authorities have quietly enrolled hundreds of thousands of incarcerated people's voice prints into large-scale biometric databases. Computer algorithms then draw on these databases to identify the voices taking part in a call and to search for other calls in which the voices of interest are detected. Some programs, like New York's, even analyze the voices of call...
|
| |
|
|
|
Researchers Fool ReCAPTCHA With Google's Own Speech-To-Text Service
Researchers at the University of Maryland have managed to trick Google's reCaptcha system by using Google's own speech-to-text service. "[The researchers] claim that their CAPTCHA-fooling method, unCaptcha, can fool Google's reCaptcha, one of the most popular CAPTCHA systems currently used by hundreds of thousands of websites, with a 90 percent success rate," reports Motherboard. From the report:
The researchers originally developed UnCaptcha in 2017, which uses Google's own free speech-to-text service to trick the system into thinking a robot is a human. It's an oroborus of bots: According to their paper, UnCaptcha downl...
|
| |
|
|
|
Taser Offers Free Body Cameras To All US Police (arstechnica.com)
An anonymous reader quotes a report from Ars Technica:
Taser, the company whose electronic stun guns have become a household name, is now offering a groundbreaking deal to all American law enforcement: free body cameras and a year's worth of access to the company's cloud storage service, Evidence.com. In addition, on Wednesday, the company also announced that it would be changing its name to "Axon" to reflect the company's flagship body camera product. Right now, Axon is the single largest vendor of body cameras in America. It vastly outsells smaller competitors, including VieVu and Digital Ally -- the company has profited $90 million from 2012 through 2016. If the move is successful, Axon could quickly crowd out its rivals...
|
| |