Research And Implementation Of Children's Auxiliary Communication And Rehabilitation Training System For Speech Communication Disorders

There are many children with speech impairment in China.According to the research,the number is about 10 million.However,the research in this area is relatively late compared with foreign countries.At present,the treatment of these children in our country is special schools,hospitals,using special equipment for one-to-one treatment.But since the base is too large,these means aren't able to benefit all the families,resulting in that a large number of children with speech impairment cannot carry out speech rehabilitation training and cannot properly communicate with people.In recent years,multi-touch technology for mobile devices and voice-related technology have blossomed in various areas,but there are few apps designed for assisted communication and rehabilitation training for children with language impairments on common mobile devices such as mobile phones and tablets.This paper introduces a design and implementation of augmentative communication and speech rehabilitation training system on Android based on current research and trend on assisted communication and rehabilitation training.First,we designed an icon-based augmentative communication system base on the research on children with communication barriers.Children can build message by tap the icons on the screen,all of which have unique voice.And a complete sentence is generated by the built-in speech synthesizer,which can provide natural sounding.A speech rehabilitation training system is built on this system.Children can use the training system to training their pronunciations of a word or a whole sentence and they get a detailed report about their pronunciation after training.We combined the augmentative communication system and the speech rehabilitation training system into one app to provide a stronger system,the trainings on the training system can give feedback to the communication system,which makes the training more attractive.To improve the input efficiency of the icon-based augmentative communication system,we introduce a non-syntactic mixed algorithm for word prediction based on the conventional N-gram model.The algorithm mixed the conventional N-gram model and a new semantic model to work on unordered message which is easily created by children with communication barriers.We also notice the needs for these children to speak directly,rather than using keyboard-like input system.We designed a new automatic speech recognition system framework which combines the speaker-dependent recognizer and the training system.Finally,we implement the whole system discussed above mainly on Android operating system with a supplemental Web interface.