What is Speech Synthesis?

Many automatic systems produce spoken messages by playing back pre-recorded words and sentences. Such systems can only reproduce the material that has already been recorded. For example, a reminder system with a pre-recorded voice might only be able to tell you that you have an appointment at 2PM because that is the only snippet of speech it has.

A speech synthesis system, on the other hand, lets the computer speak any sentence to the user. A reminder system with a fully synthesized voice will might, for example, be able to tell you that you have an appointment at 2PM at Ninewells Hospital in Dundee with consultant Dr. Smith.

Many people expect computer-generated voices to be monotone, robotic, and difficult to understand. Modern computer voices are very different. Most are easy to understand, and there are techniques for creating voices that sound very natural.

The University of Edinburgh research is creating synthetic voices that are friendly, natural, and easy to understand - even for people with hearing difficulties. Our voices cover a wide range of accents, from Received Pronunciation (BBC English) to Scottish English.