Enhancing Dysarthric Speech Intelligibility: A Review of Techniques

This paper delves into the complexities of dysarthric speech, characterized by challenges such as impaired articulation and vocal quality. To address these issues, we explore a range of techniques, including pitch manipulation, phoneme replacement, and advanced signal processing methods. Leveraging valuable datasets like Whitaker, UASpeech, TORGO, and DEED, we examine the development of robust Automatic Speech Recognition (ASR) and Audio-Visual Speech Recognition (AVSR) systems tailored for dysarthric speech. Additionally, we discuss feature space transformation methods to enhance speech intelligibility by modifying acoustic, articulatory, and auditory parameters. The integration of ASR, Text-to-Speech (TTS), and Voice Conversion (VC) technologies offers promising avenues for reconstructing intelligible and natural-sounding speech, empowering individuals with dysarthria to communicate effectively.