Considering sound as output, it can be synthesised speech, musical sounds or natural sounds. Nowadays it is mostly used as a supplementary channel for alerting and feedback purposes when the user’s visual or haptic skills need to be focussed elsewhere. Non-speech sound is universal and its meaning is learned in a very short time, similar to icons on the visual channel. However, they have to be learned and "their success is difficult to determine, since often a visual warning is at least as effective" (Dix et al. 1998, p. 561). Nevertheless the distribution of information over multiple channels will be favourable if the bundling of all leads to a disadvantageous cluttering of the medium of choice.

Researches have challenged the development of systems that are not limited to speech (listening and speaking) only but at-best use all levels of communication. This direction of research is supported by the findings from Reeves and Nass (1996). Potentially ECAs might use all five senses to gain and supply information from and to the world. But we have to consider the suitability and the nature of information they can convey.