6: Sound

6.6 Problems with the use of sound

Pre-recorded digitised speech can be included in a UI relatively easily, but generating speech is harder. One of the methods for synthesising speech is called concatenation. The idea behind concatenation is that the computer stores sentences, phrases or word segments of real human speech. New sentences are constructed by arranging words in the right order. For example, with current telephone directory enquiry systems in many countries, after having made an enquiry of a human operator, a voice says something like: ‘the number you require is 273148’. The phrase ‘the number you require is’ is smooth and flowing (having been recorded in full by a human speaker). The number itself is rather jerky and stilted, as digital recordings of the individual digits are played back one after another.

Unlike the other media that we have considered, sound has the potential to intrude upon the environment. This can be overcome with the use of headphones, but not all users choose to use headphones and they are inadvisable in some hazardous environments. It is often a good principle to allow users to change the volume, switching it off altogether if necessary.

Sound is not good at conveying detailed information, such as describing events, unless accompanied by video or still images, and it is often difficult to remember precisely. To maximise its effects, it is often best to combine sound with other media.