If a new technology is to harmoniously become part of everyday life, it must do more than just "work". It must also be accepted by its users. With regard to voice interfaces, the key to ensuring user acceptance is to take human factors into consideration in their design, in other words, issues far removed from speech recognition engines, grammars and speech synthesis. If the application is to be accepted, it will need ergonomics that work with the user.

It is quite obvious that if thousands of people call a speech portal or a call center every day, the "sound image" of the service becomes an essential component in the company's overall brand image.

The Voice User Interface designer's role is also to consider the tone and register of the language used in the service. They must also ensure that the correct amount of information is provided. In this approach, the wording of the welcome message is vitally important and is a significant part of the human factors work.

In general, 20% of the project is devoted to this area. One of the keys to a successful service is the ability to troubleshoot during an interactive dialog (by definition the user doesn't make mistakes!). Several recovery strategies can be used in the event of silence or an unexpected answer to a request issued by the system.

Besides flagging the "error", the system can repeat the question, move to a more directive dialog modality, deliver a specific help message or even transfer the user to an operator.

Ergonomics is about satisfying the user. This is achieved by providing a service which truly meets the user's needs and one they will enjoy using - A service they can relate to. This approach involves extensive research and development, the first stage of which is to study the needs, habits and culture of the target user. 

The role of the human factors specialists is to make the voice interface-based service compatible with technological capabilities, while considering the actual conditions in which the service will be used and the diversity of its potential users. Therefore, long before a system is even designed, they become observers. They look at a need - checking bank accounts on a voice server for instance - and observe how people behave, their attitudes, their vocabulary, etc. They imagine themselves in the position of the user and analyse their own reactions. Before a service is launched, they also conduct user trials in actual conditions, to help identify any malfunctions in the service or possible sources of non-acceptance.

The structure of the interactive dialogue will vary considerably depending on who the voice service is aimed at. However to be usable, the service must be intuitive for the user. The structure of the dialog will also vary considerably depending on the needs of the user. Sometimes it will give instructions, holding the user's hand and guiding them from one stage to the next. In other scenarios you can let the users learn about the system's functionality and command words by themselves In some cases you can ask the user to confirm their command words; in others, this is not necessary…

The study of users' needs is the cornerstone of any ergonomic design. In relation to human-machine dialogs, the interface designer must consider not just the variety of usage situations, but also the characteristics and heterogeneity of the target population. For instance, call frequency is of paramount importance. Someone who calls their insurance company's voice portal once a year would need a lot of guidance, whereas someone who calls a share dealing service three times a day would prefer less assistance.