The Application of Machine Learing in a Phonological Awareness System

Paczolay Dénes, Tóth László, Kocsor András és Kerekes Judit

The most important clue to the process of learning to read is the ability to separate and identify the phonemes and to associate these with its written forms. To also learn to read in a fruitful way young learners must be aware of the phonemes and must also be able to manipulate them. Many children with learning disabilities have deficiencies in their ability to process phonological information. We constructed a computer-aided training software, which uses a very effective phoneme recognizer and provides a real-time visual feedback in the form of flickering letters on calling pictures. The brightness of the letters is proportional to the speech recognizer’s output. We found we can make impressive progress in a very short training period, because the young people enjoy practicing with computers. We collected and processed a special speech database and we also chose a good combination of the examined feature extraction techniques and machine learning methods to develop the best phoneme recognizer possible.