Fact · Speech recognition
The Earnings-22 paper says modern speech recognition systems have reached superhuman word error rates on many common corpora, while lacking adequate performance on speech in the wild.
Checked against its source on October 17, 2026
In the source's own words
“Modern automatic speech recognition (ASR) systems have achieved superhuman Word Error Rate (WER) on many common corpora despite lacking adequate performance on speech in the wild.”
Earnings-22: A Practical Benchmark for Accents in the Wild, Miguel Del Rio et al., arXiv (Computation and Language), submitted to Interspeech 2022, 2022, Abstract.
Open the original- What it measures
- The authors use this observation to explain why they built a benchmark of real-world accented audio.
- What it does not show
- This is the authors' summary of the field in 2022. It does not give a specific WER for any system.