Chapter 6: User Interfaces for Mobile Media 215 enough a recognition rate until one realizes that roughly every tenth word that you speak would go unnoticed, or even worse, would be understood incorrectly in a non-predictable manner. The message would eventually get through, yet a lot of repair and repetition would be required. What is another aspect is that we cannot expect the users to start training their devices and even if they did, the domain would be limited and language-dependent. For non-constrained speech recogni- tion, there are just too many ambiguities in spoken language that need to be solved and that depend on context and common sense reason- ing, for SUIs to take off in the foreseeable future (with speech as the primary modality). There are also challenges related to mobility. Regarding use context, many environments are hostile to speech interaction. Noisy environ- ments result in recognition errors, and privacy is compromised due to overhearing, these being just two examples of context-related chal- lenges. There are also some social and psychological barriers to over- come before an SUI becomes reality for the general public. For instance, people do not wish to talk to machines. Most likely, though, this is a current trend that will eventually fade, as did “speaking to yourself” with mobile headsets a couple of years ago. Although the design of speech user interfaces shares common rules and characteristics with the graphical UIs, the whole dialogue between the user and the systems differs from the graphical user interfaces. This is partially due to the fact that audio is by nature a streaming modality (for instance, pausing does not help you to remember the last actions you performed) and hence the speech UI structure is different from GUIs. For example, with a graphical UI the user can see the state of the UI, but on speech UI the audio cannot be frozen. This implies a need for different navigation aids as it is diffi cult to operate the device when you cannot in any way sense what state the device is in, or what the available actions are. Furthermore, going through a long list of objects is slow, since the user cannot perform a quick visual scan. In addition, the current position in the list is diffi cult to present. We conclude that our focus is on graphical user interfaces, yet the advances in speech UI technologies and techniques should be fol- lowed closely. 6.3.2 Graphical User Interfaces To understand graphical UIs for mobile devices, let us look at graphical WIMP user interfaces that have dominated desktop computing for more than two decades. The acronym WIMP highlights the key char- acteristics of the user interface: windows contain information, menus