Ensure that the spoken output is understandable
Users will have difficulty understanding spoken output that is not clearly spoken in their own language or dialect. Responsiveness is also an issue. If speech is delayed and lags behind the event being spoken, users may become frustrated or misunderstand what is being referred to, leading to errors.
Directions and techniques
Synchronise spoken output with events or focus (high priority)
Spoken output should be synchronised with the event that is being described, such as the change of focus from one item to another when the user presses a button on the remote control. This will ensure that that spoken output is perceived as belonging together with the event being spoken.
Having sufficient processing power to avoid time lags is important in this regard so the requirements will need to be taken into account in the early stages of designing products with spoken output.
Use an appropriate language (high priority)
Text-to-speech software should be designed to speak the language of the receiver user interface. A speech synthesizer designed for one language will not read another language with the correct accent and may therefore be unintelligible.
It is preferable to use a language or dialect local to the users of the equipment.
Make sure the information itself does not rely on vision to be understood
Do not write information and instructions in the user interface in a way that relies on the perception of shape, size, colour or position. For example, the instruction to choose from the options on the right. Will be difficult to follow unless the user interface provides a simple and intuitive way for moving left and right and the list of options is the only thing on the right. Even if this is the case, it is still likely to present problems for some users who will wonder, as they move to the right, have I reached the options yet.
If instructions like this exist within the user interface, the text to speech engine should substitute a more explicit alternative description.
Text to speech engines should avoid stuttering or pauses in speech. This can occur if the speech processor is too slow to keep up with what it is being asked to do, so the processor requirements will need to be taken into account in the early stages of designing products with spoken output.