Academics creating the first speech recognition system in Maltese have reached out to the community to provide them with as many diverse voice samples as possible.

Senior lecturer of language technology Albert Gatt told Times of Malta that the new technology dependent on these data would be limited without them.

“Speech is the most natural way to communicate and people are being deprived of some useful functionality that would make interaction with these systems and machines much smoother,” he said.

This technology underlies the facility to give voice commands to your phone using software like Siri, Gatt explained.

It would enable people who don’t have the ability to type – somebody who has a disability for example – to send messages.

Other uses for this technology would be to cut down on phone line queues as people would be relayed to the right places using voice recognition.

People wishing to participate in the project simply have to follow the instructions on the Facebook page Maltese Automatic Speech Recognition and read out sentences, which are then recorded.

They also have the ability to check how others have read out sentences and to validate them, giving their opinion as to whether they have read them out properly or not.

It was important to have a diverse sector of Maltese society participate in this project, Gatt continued.

People from different age groups and social backgrounds spoke with slightly different dialects and had different accents.

“If we only collected voice samples from white 40-year-old males then the voice recognition most likely wouldn’t be able to recognise a 16-year-old female,” he pointed out.

Since launching their crowdsourcing Facebook page last week, the team behind the project have managed to collect four hours of data and they hope to have an initial prototype system set up by February 2021.

Collecting the data was also of scientific interest, Gatt underlined, seeing how a lot of artificial intelligence works with these fast data sets.

The project, which has been funded by the Research Excellence Award, has been running for nine months but researchers have spent most of this time setting up the systems.