Text-to-speech technology has evolved in such a way that it allows now to let people reproduce their voice in a rather easy way. The recording is done by means of an internet application. It takes you about 2.5 hours to record about 1500 sentences. You can do this in different sessions. The aims of this paper is to demonstrate the possibilities of voice banking.
In order to evaluate this 'new' text-to-speech technology we did a try-out and reproduced one of our voices and implemented the result in one of our SGDs and presented this to the social network of the person who's voice was 'reproduced'. The use of the 'wearable voice' allowed us to make the experience even more 'personal' as the voice didn't come out of the SGD but from a little wearable device.
The evaluation process is stil ongoing but the first findings are exciting. Relatives, friends really recognise the voice as being from the person . The identity that a voice carries was in fact still there. This is not the case with 'standard' text-to-speech voices. And in fact there are not a lot of text-to-speech voices available (Nuance, Acapela, Microsoft...). So a lot of AAC users speak with the same voice.
This could bring AAC to another level. People who know that there is a chance of losing their voice could make a recording of their voice.