May 19, 2020

Microsoft has developed an innovative speech recognition system

Microsoft
Computational Network Toolkit
GitHub
WER
Catherine Rowell
2 min
Microsoft has developed an innovative speech recognition system

Microsoft has recently announced that they have created speech recognition technology which has enabled computers to encompass the same capacity to identify and recognise words to the same capability as the everyday person, minimising the word error rate (WER) to solely 5.9 percent.

Researchers have attributed this success not only to the company’s dedication in creating a sufficient system, but the use of the latest neural network technology has been embedded throughout the system, enabling the system to recognise words and patterns.

Microsoft’s research goes back over twenty years, and paves the way for more intelligent technologies and future robotics, impacting on future business models and consumer goods with the introduction of sophisticated speech recognition and development of assistants such as Cortana.

The technology giant has utilised their Computational Network Toolkit within the creation of the system, which has become available on GitHub.

However, there is still further development that needs to be made to ensure the technology is sufficient for real-life situations, with the notion of background noise, with renovation works, increased noise on highways, shopping centres, in addition to incorporating and ensuring different regional accents are understood, etc.

Microsoft will also be working on refining the system in order for not only to record words and speech, but to understand what the information means and recognise different individuals/voices within discussions.

For further information: https://mspoweruser.com/microsoft-research-breaks-human-speech-recognition-barrier/

Follow @BizReviewUSA and @NellWalkerMG

Read the October issue of Business Review USA & Canada here

Share article

Jun 16, 2021

Dr Peng Wei: Designing the Future of Autonomous Aircraft

NASA
Sustainability
IATA
Airbus
3 min
NASA has announced that it will fund a new project, headed by Dr Peng Wei, to develop safety management systems for autonomous electric aircraft

Air traffic is expected to double by 2037. According to the International Air Transport Association (IATA), the world will need 37,000+ new passenger and freight aircraft, and more than half a million new pilots—unless we come up with another solution. Right now, a George Washington University School of Engineering and Applied Science professor, Dr Peng Wei, is starting to research autonomous electric aircraft design. 

 

NASA will fund the research, which will study how to minimise risks for electric vertical take-off and landing (eVTOL). As Airbus states: ‘Autonomous technologies also have the potential to improve air traffic management, enhance sustainability performance and further improve aircraft safety’. 

 

Who is Dr Wei? 

An assistant professor of Mechanical and Aerospace Engineering, Dr Wei has researched aircraft control, optimisation, and AI and ML applications in aviation. Over the next three years, he’ll lead the US$2.5mn NASA grant project in collaboration with researchers from Vanderbilt, the University of Texas at Austin, and MIT’s Lincoln Lab

 

Why is His Research Important? 

Even though the wide adoption of self-piloting cars, much less aircraft, is still far down the road, technologies that Dr Wei and his colleagues are researching will form the commercial transport of the future. But aviation manufacturers, in order to produce autonomous aircraft, will have to meet extremely high safety standards. 

 

‘The key challenge for self-piloting capabilities is how the system reacts to unforeseen events’, said Arne Stoschek, Wayfinder Project Executive at Acubed. ‘That’s the big jump from automated to autonomous’. In the air, AI-piloted aircraft will have to manoeuvre around adverse weather conditions, such as wind and storms, and other high-altitude risks, such as GPS hacking, cyberattacks, and aircraft degradation. And the stakes are high.

 

‘If a machine learning algorithm makes a mistake in Facebook, TikTok, Netflix —that doesn't matter too much because I was just recommended a video or movie I don't like’, Dr Wei said. ‘But if a machine learning algorithm mistake happens in a safety-critical application, such as aviation or in autonomous driving, people may have accidents. There may be fatal results’. 

 

What Are His Other Projects? 

In addition to the new NASA research, Dr Wei has been awarded three other grants to pursue AI-piloted aircraft: 

 

 

Research like NASA and Dr Wei’s three-year programme will help improve how AI reacts and adapts to challenging air conditions. In coming years, autonomous aircraft will likely take off slowly, starting with small package delivery, then upgraded drones, and finally commercialised aircraft. But congestion issues will worsen until autonomous aircraft are the best alternative. 


According to BBC Future, by 2030, commuters will spend nearly 100 hours a year in Los Angeles and Moscow traffic jams, and 43 cities will be home to more than 10 million people. The final verdict? Bring on the AI-operated transit.

Share article