Automatic Speech Recognition – State of the Art 2025 and the role of human intervention

Scheduled at 9:00 am in Matchless on Thursday, November 20.

Speaker(s)

• Daniel Goerz, CEO, Habitat Learn Inc

Session Details

• Length of Session: 1-hr
• Format: Lecture
• Expertise Level: Beginner
• Type of session: General Conference

Summary

Explore how speech recognition is reshaping accessibility in higher education through closed and live captioning. This session examines the strengths and limitations of automatic speech recognition (ASR) technology and its application in real-world classrooms to support students with learning challenges. The session will examine bias, room acoustics and the importance of human intervention. Attendees will gain practical strategies for thoughtfully integrating ASR into student workflows.

Abstract

This session offers a comprehensive, research-backed look into the evolving role of Automatic Speech Recognition (ASR) and human captioning in higher education accessibility. As institutions increasingly adopt ASR for closed and live captioning, it’s critical to understand both its potential and its persistent shortcomings. Drawing on Habitat Learn’s extensive experience as a leading captioning provider in Canada, and its ongoing research into bias in ASR—covering age, gender, race, and linguistic diversity—this session examines the systemic challenges of engines trained predominantly on white, male, mid-range voices.

We’ll explore how ASR accuracy varies by accent, tone, and audio environment, and how these factors shape equitable access for diverse learners. The discussion will highlight how human captioners, supported by advanced software productivity tools, can significantly improve real-time accuracy and comprehension in live captioning scenarios—the most technically and cognitively demanding context for accessibility.

Attendees will learn where ASR can add value and where human expertise remains indispensable. The session will feature practical workflows showing how hybrid human-plus-AI captioning can achieve both efficiency and inclusion. This session is especially valuable for accessibility teams seeking to balance automation, compliance, and student equity across learning environments.

Keypoints

1. Research into bias aspects of ASR can inform judgement on when to use human intervention
2. Understanding student workflows can deliver better solutions for supporting their learning
3. Most ASR solutions do not reach quality and accuracy thresholds required in higher Education

Disability Areas

Cognitive/Learning, Deaf/Hard of Hearing

Topic Areas

Artificial Intelligence, Assistive Technology, Captioning/Transcription, Uncategorized

Speaker Bio(s)

Daniel Goerz

Note taker, Learning Strategist, Exam Proctor Alternate Format Specialist (Kurzweil), Assistive Technologist Founded Note Taking Express(renamed Habitat Learn Inc) in 2014 to better serve diverse learners Leads Messenger Pigeon, a platform built with input from students and accessibility teams Advocates for inclusive, human-centred technology Believes technology is a tool, not a replacement, for thoughtful, community- driven support