In order to develop Translate we need to collect as much data as possible. We are inviting you to help us gather short recordings of the vocalisations made by people who are non-verbal.
We will use these anonymised sound samples to train an algorithm so that it can provide an accurate of up to 10 emotional states.
The process is quick and simple. A carer, who knows the non-verbal person well, is asked to record 5-15 seconds of his/her verbal communication, then to touch one of the 10 emoji’s that represent various emotions. You can do this as often as your time permits.