In the Nowrin and Vertanen study, slowing the read-back audio only where the recognizer was unsure led to a 12 percent relative increase in how well listeners detected errors, compared with uniformly slowing the audio.

Checked against its source on October 16, 2026

In the source's own words

“selectively slowing down the audio when the recognizer exhibited uncertainty led to a 12% relative increase in participants' ability to detect errors compared to uniformly slowing the audio”

Using Confidence Scores to Improve Eyes-free Detection of Speech Recognition Errors, Sadia Nowrin and Keith Vertanen, arXiv, 2024, Abstract.

Open the original

What it measures: The comparison is against slowing the whole read-back audio by the same amount. The method uses the recognizer's own confidence scores.
What it does not show: A relative increase, not a rise from one share to another. The study is a lab experiment, and no product currently does this for you.

Source

Using Confidence Scores to Improve Eyes-free Detection of Speech Recognition ErrorsPreprint · Sadia Nowrin and Keith Vertanen, arXiv, 2024

Cited on

Dictation for Blind and Visually Impaired PeopleGuide