Fact · Dictation
Radford and colleagues report that, compared with humans, their Whisper speech recognition models approach human accuracy and robustness.
Checked against its source on October 16, 2026
In the source's own words
“When compared to humans, the models approach their accuracy and robustness.”
Robust Speech Recognition via Large-Scale Weak Supervision, Alec Radford et al., arXiv (eess.AS), 2022, Abstract.
Open the original- What it measures
- The models were trained on 680,000 hours of audio and tested on standard benchmarks without fine-tuning.
- What it does not show
- A benchmark comparison from a 2022 preprint. It does not measure your microphone, accent or room, and the human comparison depends on the test set.
Source
Cited on
- Dictation and Voice Typing: How to Talk Instead of TypeTopic
- Dictation Software: How to Choose the Right AppTopic
- Speech Recognition: What It Is and How Well It WorksTopic
- Transcription: How to Turn Audio and Video Into TextTopic
- Dictation Languages, Accents and Voice TranslationTopic
- Best Dictation Software and Apps, Compared by NeedGuide
- What Is Whisper AI? OpenAI's Speech Model ExplainedGuide
- Free Dictation Software: The Best Free OptionsGuide
- Dictation vs Typing: When Speaking Beats the KeyboardGuide