Synthetic speech promises to ungarble PA announcements in noisy places

EDINBURGH, SCOTLAND—We’ve all heard PA announcements that sound something like this: 

ONNAWAWA Airlines Flight seven-WAWAWA-ty-four will depart from Gate sixty GAWAGWA at WAWAWA forty-five.”

Or, “This subway train has been delayed because of WAWAOMOMONWA.  Passengers should not ONHONHOHN until WAWAWA. Thank you for your patience.”

But now, according to a report from the University of Edinburgh in Scotland, scientists there have come up with a way to make public announcements in noisy places not only audible, but understandable as well.

To improve the state of the art in synthetic voice technology, the scientists studied how listeners perceived speech. They conducted tests to pinpoint the components of speech that are most easily heard by people in a noisy place.

The Edinburgh group knew that in loud settings such as railroad stations, airports, and sports arenas, listeners pay most attention to the parts of speech that are easiest to hear, and use those to decipher what is being said. Therefore, the scientists conducted tests to pinpoint the particular components of speech that are most easily heard by people in a noisy place.

They used a mathematical computer program to analyze spoken words and enhance the sounds that are most helpful to listeners in understanding what is being said. They then developed software that can alter speech before it is broadcast over speakers, making it more audible amid background noise.

Tests showed that this synthetic speech was much easier to understand than natural speech. In some cases, the improvement was the equivalent of lowering the background noise by five decibels. Scientists say that the techniques used in this study could also be used to improve smartphone voices.

Cassia Valentini Botinhao, PhD, of the University of Edinburgh’s School of Informatics, who conducted the study, said, “Noisy environments make it difficult to understand what is being said and simply making speech louder isn’t the smartest solution. Our findings could offer an alternative, by making speech more intelligible without turning up the volume.”

The study, which was carried out with scientists from Greece, Spain and Sweden and funded by the European Union, was presented in August at the Interspeech 2013 Conference in Lyon, France.

SHARE THIS!