Preliminary Investigations in Automatic Recognition of English Sentences Uttered by Italian Children

This paper reports on a initial research activity in the area of non-native children’s speech recognition that was carried out by exploiting two children databases, one consisting of speech collected from native English children, the other one consisting of English sentences read by Italian learners of English in the same age range of the native speakers. By exploiting the corpus of native speech a baseline speech recognizer was trained for British English. Recognition results achieved for native speakers were compared with those achieved for Italian children uttering the same English texts. Word error rates achieved for Italian children were 100%600% higher than those achieved for English children of the same age. Adapting baseline acoustic models by using a small amount of nonnative speech from a group of Italian learners of English, resulted in a great improvement in recognition performance on non-native speech.