A Study of Artificial Intelligence-Driven Educational Intervention Strategies for Students with Special Educational Needs

This paper first introduces the intervention path of artificial intelligence in special education. It then takes the deaf and hard-of-hearing people as the research object, adopts artificial intelligence technology, digitizes the speech signal, performs sound-rhyme syncopation, and constructs an interactive speech evaluation system using the HMM model as the evaluation basis. Finally, this paper discusses the effect of an interactive speech recognition model on special education interventions and describes the future development trend of combining artificial intelligence and special education. The results show that the subjects' passive response to basic interactive speech in the baseline period showed a stable state, the interactive speech ability in the intervention period showed stable progress, and the subjects' ability in the maintenance period remained at a high level. In the three periods, the posttest error rates of the experimental group testers on semantic recognition were 40%, 30%, and 10%, respectively, whereas the control group was 80%, 70%, and 65%, respectively, which shows that the semantic network training affected the improvement of the speech recognition ability of the 10 subjects. The mean values of the average error rate of the three stages of the experimental group were 7.53%, 2.44%, and 2.07%, and their average recognition times were 1566.73 seconds, 1051.30 seconds, and 463.55 seconds, respectively. The three-stage semantic recognition training significantly improves the subjects' semantic recognition ability and reduces the recognition time. The integration of artificial intelligence and special education is critical for special education groups' education and healing. © 2024 Dongxia Mo et al., published by Sciendo.