Dev_Software
Until we have a physical prototype software development will be limited to client-side applications.
For the braille input devices I would like to add a 3 button input to the device to enable word completion to the user. Beginning from a state of standard input the first button will audibly read the most likely word the user means to input. The other two buttons will be used to accept or decline the proposed word. If the word is declined then the next word will be read for the user. If the word is accepted the sequence is terminated and the word will be sent to the braille output for verification. If the word is neither accepted nor declined within a predefined timeout window the sequence will terminate with no change to the input, this same action will occur if the original "start" button is pressed at any point in the sequence.
Below is a flowchart documenting the interface I plan to develop from the text prediction to the braille output and the text to audio output. More research must be done on existing text prediction software in order to develop an algorithm interfacing the braille input to the text prediction.
In short, the next step is to research what input methods are used by popular text prediction software and develop an algorithm to adapt. Ideally I would like one with an internal buffer that takes one character at a time and a terminator, but I also have preexisting ideas to handle algorithms that take an entire word and return recommendations. I can accommodate for other methods of input but either of these two would probably best fit our needs.
The specific needs towards text to audio software are much less strict than those on the text prediction, the majority of performance will come from the text prediction.