The evolution of mobile devices, such as tablets or touch phones, has led to the disappearance of the physical keyboard. Text input on touch media is now done directly using on-screen keyboards. The problem is that the touch screen has no relief and therefore cannot provide any information to blind and partially sighted people about the position of their fingers relative to the keys on the keyboard.
The objective of the project is therefore to propose a specific input system that allows these visually impaired users to enter text as quickly and with as much precision as they are used to doing with a physical keyboard.
Beyond the objective of optimizing data entry performance (i.e. data entry speed and precision), the challenge of the project is to offer a solution that requires little or no learning to users. For this, one of the design constraints will be to offer interaction techniques compatible with the AZERTY keyboard, the layout of the keys is known to all.
The other challenge is to study the usability of dynamic systems (such as prediction systems) to improve capture. This type of system is frequently used by sighted people, but difficult to use by blind people because of the dynamic aspects that are difficult to present.
Currently, two main trends coexist: specialized systems, which respond to current disability constraints, but which are often too complex to use for novice users; and augmented keyboards, which are often summarized as a voice addition on an already existing keyboard, without offering a real interaction method that compensates for the delay induced by voice synthesis, or the additional cognitive load attributed to the user.
DUCK (deductive keyboard) is based on the AZERTY and QWERTY provisions. These provisions are the most used, which means that most people are used to them. The choice of DUCK is in the simplification of the interaction for typing letters. The main objective is to get rid of the long search for each character on the keyboard using speech synthesis. At the same time, DUCK seeks to compensate for the lack of precision linked to mobile keyboards and small ones.
DUCK is a full screen keyboard, which prevents the user from invoking submenus by mistake. The user first explores the keyboard to find the first letter of the word sought. Each finger movement is linked to a voice feedback, allowing the user to locate a letter on the keyboard. Once found, he releases the key, the corresponding character is selected. To type the remaining letters of the word, the user does not have to explore the keyboard again. He just needs to press where he thinks the keys are, depending on his memory representation of the AZERTY or QWERTY keyboard. The user signals the end of the word by pressing on the screen with two fingers. The deduction algorithm then offers the best matches to the user, by spelling the entire words to avoid any confusion.