Therapy Box has invested heavily in developing specific speech recognition software that works with children and a variety of speakers with speech impediments. We are currently leading the field in this area of research as most speech recognition tools on the market are currently only optimised for adult speakers, without any speech difficulties.
Mainstream speech recognition does not work with dysarthric speakers, leaving a significant portion of people with disabilities unable to access their device in a way that should be both quick and simple. We have put this advance technology into one of our newest apps, VocaTempo.
We also create and develop a range of speech recognition engines for multiple different uses, suited to a range of user groups. Our focus on speech recognition of children’s voices is unique. Articulation Forrest is able to not only recognise what the child is trying to say, it is also able to recognise if they have pronounced the word correctly.
In our VocaTempo project, we worked with the University of Sheffield and Barnsley Hospital to develop an app which can be trained to recognise dysarthric speech. This app makes it easier for people with disabilities to access and control their devices. Dysarthric speech recognition technology has many potential applications in communication, accessibility and articulation rehabilitation, which we are committed to exploring in the coming months and years.
Using speech recognition, acoustic analysis and advanced machine learning, ATLAS has automated the process of screening for developmental language disorder (DLD). The system uses a self training engine that adapts to new information and “learns” to become more accurate over time and use. Focussed on child speech, the novel speech recognition requirements extend our capabilities to this important area.