Scottish university aims to spark research into new voice recognition technologies

A Scottish university has revealed groundbreaking speech recognition analysis.

The University of Glasgow has analysed muscle movement involved in speech to fill the gap in public data available to support new voice recognition technologies. 

The data is free to use, with the hope that it will progress treatments for those suffering from communication difficulties – a condition which affects more than 275,000 children in Scotland. 

The figure increases to almost 1.5 million across the UK. 

Professor Muhammad Imran, co-author of the paper, said: “Contactless sensing has huge potential for improving speech recognition and creating new applications in communications, healthcare and digital security.  

“We’re keen to explore in our own research group here at the University of Glasgow how we can build on previous breakthroughs in lip-reading using multi-modal sensors and find new uses everywhere from homes to hospitals.” 

The speech impediment crisis, practitioners have said, was exacerbated by Covid. Almost 90 per cent of them reported they had seen an "increase” or “significant increase” in the number of children with communication difficulties after the pandemic, according to a survey by the Royal College of Speech and Language Therapists (RCSLT) and Early Years Scotland. 

Future technologies could help people suffering from speech impairment or voice loss by using sensors to read their lips and facial movements and give them access to a synthesised voice.  

Professor Qammer Abbasi, author of the research paper, said: “What we set out to do in collecting the RVTALL dataset was create a much more complete set of analyses of the visible and invisible processes which create speech to enable new research breakthroughs, and we’re pleased that we’re now able to share it.” 

To build the data set, researchers analysed speech via a range of wireless sensing devices. 

Twenty volunteers were asked to speak a series of vowel sounds, single words and entire sentences. 

As participants spoke, radar technologies helped researchers image facial, tongue and larynx movements. Meanwhile, a laser system scanned skin vibrations and a camera captured deformations in the participant’s mouths. 

The data was then validated by signal processing and machine learning techniques, building a detailed picture of the physical movements which allow people to form sounds. 

It could also make banking transactions safer. It could analyse users’ unique facial movements before they unlock information such as their fingerprints.  

The paper, titled ‘A comprehensive multimodal dataset for contactless lip reading and acoustic analysis’ has been published in the journal Scientific Data.  

The Engineering and Physical Sciences Research Council and the Royal Society of Edinburgh funded the innovative research.