What is synthetic speech?
Synthetic speech is artificial human speech, which is produced by a computer. There are a number of different software applications through which this process can be achieved.
Some speech synthesisers use pre-recorded human speech and fit words together to form sentences (this is most often used in applications with a limited vocabulary, such as a talking clock). Other synthesisers are more complex in that they fit together tiny portions of speech (sounds) to form words and sentences. Using this method, a synthesiser is able to produce an unlimited vocabulary, and can therefore read aloud any text input. This is known as text-to-speech synthesis.
How is synthetic speech used?
Many people will have experienced synthetic speech, possibly without even knowing it. Common applications used in the wider world include railway station announcements and telephone systems. Synthetic speech is also available on products, such as mobile phones or music players which can "read" out menu systems or portions of text.
How is synthetic speech used by blind and partially sighted people?
As many blind and partially sighted people access information through audio, synthetic speech has a wide range of applications. These range from simple talking products (such as clocks and thermometers), to reading materials (such as books), to access technology such as screen readers which allow blind and partially sighted people to access computers.
Screen reading software on computers is a key use of synthetic speech for blind and partially sighted people. This uses a text-to-speech engine, where the computer reads aloud the text displayed on the computer screen.
What are the benefits of synthetic speech?
-
Speed of access: recording information via human narrators is time consuming. Synthetic speech can be produced very quickly from an electronic text file, meaning synthetic speech may enable quicker access to information
-
Cost: in the long term, synthetic speech may be cheaper to produce than human narrated audio. Whilst the initial outlay for synthesis technology may be high, reduction in the time spent creating books could be a saving
-
Confidentiality: for confidential information - such as financial statements or medical information - synthetic speech offers a means of access without someone else (a human narrator, friend or family member) having to relay the information to you
-
Speed of delivery: synthetic voices are designed to be clear at speed, and can be sped up beyond the rate at which a human could speak.
What are the disadvantages of synthetic speech?
-
Articulation: while technology is always advancing, it is fair to say that even the best synthetic voices don't sound quite human. Some users complain about robotic voices, poor pronounciation and strange intonation.
-
Intelligence: similarly, synthetic voices are not human, and cannot pick up on the context of the information they are portraying. When used for audio books, some users dislike the lack of human input, such as with humour, putting on accents or voices for different characters or building suspense in their voice. The importance of such features is likely to vary depending on what the synthetic speech is being used for.
-
Attention: some users find synthetic speech more difficult to listen to and find they have to pay more attention to it compared to human speech. This may be because of small anomalies in pronunciation etc. This could affect reading performance, although it is commonly reported that people get used to listening to synthetic speech over time.
Further information
For more information on how blind and partially sighted people use synthetic speech (published November 2008):
For research into how talking book readers feel about the idea of talking books in synthetic speech (published May 2009):
For guidance on how to evaluate synthetic voices for use in products or services (published January 2010):