Speech-to-text and gesture recognition tools for non-verbal autistic accessibility

Many autistic people have long-term problems with communication, including poor proficiency in spoken language and the use of non-verbal language. These challenges are core impediments to quality interaction and independent living at clinical, educational and social levels. Although the traditional Augmentative and Alternative Communication (AAC) based, are fundamental, they do not usually have the ability to maintain real-time personalization and responsiveness that is needed to support fluent, high-fidelity interaction. The following review discusses how the Artificial Intelligence (AI) platforms, including speech-to-text (STT) and gesture recognition, have become sustainable accessibility tools to the non-verbal population with autism. A systematic review of recent literature assesses key themes that define successful deployment: precision, emphasis on algorithmic resilience to non-stereotypical vocalizations and idiosyncratic gestural communication; the requirement of sensory alignment in interface design to accommodate neurological variations; the imperative of inclusivity, particularly in relation to access and representational datasets; and overarching ethical concerns, including data privacy, possible sensory load, and algorithmic bias. This discussion finds that although AI-driven accessibility has a huge promise to promote autonomy and optimize the speed of interaction, its implementation demands a high-level of clinical alignment, interdisciplinary and commitment to neurodiversity-informed design values.