Study could improve automated booking phonelines


By Cara Sulieman

THEY ARE supposed to make our lives easier, but end up trying to book us in to see a film at a cinema miles away.

Now scientists at Edinburgh University are hoping they can improve automatic phone booking lines after pinpointing the most common errors made by the computers.

Companies use the systems to cut out a call centre – using computers that can understand what customers are saying to book cinema tickets and taxis.

But it doesn’t always work, and can lead to a frustrating conversation with a computer.


Boffins carried out a study to find out what confuses the computer, and are hoping it can be used to refine existing systems.

They found that speech peppered with ‘umm’ and ‘err’ was the most likely to cause problems and that men are more likely to be misunderstood because they ‘umm’ and ‘err’ more than women.

They also found that the first word in a phrase is the least likely to be picked up by the system – either because the computer can’t put it in context or because the caller inhales just before talking.

The researchers recorded phone calls and fed the conversation into a speech recognition system to see how much it understood.

The computers couldn’t cope with words that sounded similar, especially when the word wasn’t said properly.

“Very frustrating”

Changes in pitch, tone and speed also confused the system.

Dr Sharon Goldwater, of the University of Edinburgh’s School of Informatics said the she hoped the study would help make booking system more efficient.

She said: “Everyone has had some of these very frustrating experiences with automated phone lines.

“Voices vary from one person to the next and it is challenging to design a computer system that can understand lots of different voices.

“We hope that by closely studying how people speak and how machines process this, we can help create better systems that will be simple and efficient for people to use.”

The study was published in the Speech Communication journal and was sponsored by the US Office of Naval Research.