Interaction between people with dysarthria and speech recognition systems: A review

Area of Science:

• Speech and Language Technology
• Human-Computer Interaction
• Assistive Technology

Background:

• Automatic Speech Recognition (ASR) systems have advanced rapidly, but their accessibility for individuals with speech impediments, such as dysarthria, is not well-documented.
• The period reviewed (2011-2020) covers the launch and growth of commercial voice assistants like Siri.
• Existing research often overlooks the specific interaction challenges faced by users with dysarthria.

Approach:

• A systematic literature review was conducted, focusing on studies published between 2011 and 2020.
• Articles were selected from scientific databases and journals based on predefined inclusion criteria.
• The review utilized an interaction framework encompassing user, input, system, and output elements.

Key Points:

• A total of 36 studies met the inclusion criteria, with Word Error Rate (WER) being a common evaluation metric.
• Significant interaction challenges persist for individuals with dysarthria when using current ASR technologies.
• The overall interaction process between users with dysarthria and ASR systems is not fully understood.

Conclusions:

• Current ASR systems, despite recent technological progress, continue to present usability barriers for individuals with dysarthria.
• A deeper understanding of the complete user-system interaction is crucial for enhancing ASR accessibility.
• Future research should focus on elucidating ASR advancements to better support users with speech impediments.