Fact · Speech recognition
20.1percent average word error rate
In a 2024 study of eleven speech recognition services on university lecture recordings, Google's average word error rate on English was 20.1 percent.
Checked against its source on October 17, 2026
In the source's own words
“Google 20.1 23.6 28.1 18.1 IBM 11.2 13.2 17.3 20.6”
Measuring the Accuracy of Automatic Speech Recognition Solutions, Korbinian Kuhn et al., arXiv (preprint), 2024, Table 4, English column.
Open the original- What it measures
- The English column is the average over the lecture recordings in batch mode without a custom vocabulary. The next column, LibriSpeech, gave Google 23.6 percent.
- What it does not show
- Lecture audio, one test. The paper describes Google's service as it was when the study ran, and cloud models change, so newer Google models may score differently.
Source
Used to calculate
Cited on
- Zoom, Teams and Google Meet TranscriptionGuide
- Transcription Software: How to Choose OneGuide
- Speech Recognition in Python: Libraries ComparedGuide
- How do you transcribe a lecture recording?Question
- How do you transcribe a Zoom audio or video recording?Question
- Is Whisper more accurate than Google Speech-to-Text?Question