Automatic Speech Recognition for Speech Assessment of Persian Preschool Children

Preschool evaluation is crucial because it gives teachers and parents influential knowledge about a children's growth and development. The coronavirus pandemic has highlighted the necessity of online assessment for preschool children. One of the areas that should be tested is the ability to speak. Because of the differences between children's and adults' voices, employing Automatic Speech Recognition(ASR) systems is difficult since they are pre-trained on adults' voices. We constructed an ASR for our cognitive test system to solve this issue using the Wav2Vec 2.0 model with a new pre-training objective called Random Frequency Pitch(RFP). In addition, we used our new dataset to fine-tune our model for Meaningless Words(MW) and Rapid Automatic Naming(RAN) tests. Our new approach reaches a Word Error Rate(WER) of 6.45 on the Persian section of the CommonVoice dataset. Furthermore, our novel methodology produces positive outcomes in zero- and few-shot scenarios.