Fact · Transcription
2.15times the headset word error rate
With the same model, Whisper Large V2 had about 2.15 times the word error rate on the single distant microphone recording of AMI meetings as on the headset recording.
Checked against its source on October 17, 2026
- What it measures
- The ratio divides the single distant microphone word error rate by the headset word error rate for the same model, Whisper Large V2, on the AMI meeting corpus. It shows the size of the gap between the two recording conditions, not an error rate for any particular meeting.
- How it is calculated
- 36.4 / 16.9 = 2.15
Calculated from
Cited on
- Meeting Transcription: How It Works and What to UseGuide
- Speech Recognition and Accents: Accuracy and FixesGuide
- Are AI transcription devices and voice recorders worth it?Question
- Can you transcribe a meeting without a bot joining?Question
- How do you choose the microphone Windows dictation uses?Question
- How do you transcribe an in-person meeting?Question
- Why doesn't voice recognition understand my accent?Question