This week’s list of top data news highlights covers October 1, 2022 to October 7, 2022 and includes articles on using a supercomputer to determine the Moon’s origin and improving speech recognition models.
Google has created an AI system that can generate speech and music from audio snippets. The team used tokenized audio recordings featuring variables that affect the perception of sound, such as syntax, rhythms, phonetics, or recording conditions, to train the system to recognize and extend sound patterns in submitted audio prompts.
Researchers at Curtin University in Australia have used a supercomputer to simulate tectonic movements and determine the location of the world’s next supercontinent. The team found that the Australian and North American continents will likely collide with Asia within the next 200 to 300 million years to form a supercontinent called Amasia.
Researchers at Houston Methodist Hospital in Texas have created an AI system that can predict hospitalization outcomes, defined as the length of stays or the locations of post-discharge care, for geriatric patients with dementia. The team trained the system with medical and demographic data from 8,407 patients. In tests, the system achieved 95.6 percent accuracy.
Researchers at Durham University in the United Kingdom have used a supercomputer to simulate the creation of the Moon. Previously, the leading theory for the Moon’s creation hypothesized that the Moon was a result of gradual accumulation of debris from an impact between the Earth and another celestial body. However, the researchers’ simulation suggests that the impact immediately placed the Moon into orbit.
Apple, Amazon, Microsoft, Meta, Google, and researchers at the University of Illinois have partnered with nonprofit research organizations to launch the Speech Accessibility Project, a research initiative to improve voice recognition models. The team aims to create voice recognition models for people with diverse speech patterns or disabilities that affect speech.
Researchers at the University of Michigan and University of Purdue in Pennsylvania have used a supercomputer to simulate tsunami activity after the Chicxulub impact, or the asteroid impact that killed the dinosaurs. The team found that the impact caused a tsunami that reached nearly three miles high and traveled for 140 miles in every direction.
The World Health Organization has updated its AI-powered virtual health worker Florence. Previously, Florence primarily provided advice on how to quit smoking. The new version can share health advice on more topics, including mental health, COVID-19 vaccines, and nutrition.
Researchers at Scotland’s National Robotarium have created a smart helmet that can help firefighters locate victims in smoke-filled rooms. The helmet contains sensors, thermal cameras, and radar technology that collect data on firefighters’ surroundings and an AI system that uses the data to identify victims or other firefighters.
The Smithsonian has launched an app-based walking tour through the Anacostia neighborhood in Washington, D.C. The app, called Before the Bulldozers, uses augmented reality to display information about the neighborhood’s history and residents’ displacement in the 20th century.
Researchers at St. George’s, University of London have created an AI system that can predict a patient’s chance of cardiovascular disease, cardiovascular death, and stroke from retinal images within 60 seconds. The team developed the system with retinal images from 88,052 patients.