Glossary
Diarization Error Rate (DER) is the standard accuracy metric for speaker diarization. Unlike WER which measures word-level errors, DER measures the percentage of audio time where the diarization system gets the speaker wrong. Lower is better. State-of-the-art systems achieve 5-15% DER on conversational benchmarks; production diarization on a clean 2-speaker podcast can drop to 3-5%.
What DER counts
DER = (Missed speech + False alarm + Speaker confusion) / Total speech time. Missed speech: the system marked it as silence when there was actually speech. False alarm: the system thought there was speech in a silence region. Speaker confusion: the system labeled speech as Speaker A when it was actually Speaker B. The three components are summed and divided by the total reference speech duration.
Benchmark numbers (2026)
DIHARD III (hard, real-world): SOTA ~15-18% DER. VoxConverse: SOTA ~4-5%. AMI Meeting Corpus: SOTA ~12-15%. Production 2-speaker podcasts (clean recording, no overlap): typically 3-7% DER. Hard cases — overlapping speech, voices that sound alike, very short utterances — push DER up significantly. Most academic benchmarks include a 250 ms forgiveness collar around speaker boundaries to be lenient on short alignment errors.
DER vs WER
They measure orthogonal things. WER measures whether the words are right; DER measures whether the speaker labels are right. You can have 0% WER (perfect transcript) and 50% DER (every speaker label is wrong). Conversely, 50% WER and 0% DER means you got the speakers right but mistranscribed half the words. Production transcription cares about both — we report each separately when measuring quality.
In practice
A 60-minute podcast with 2 hosts. Reference: each host talks 30 minutes. Our system labels 28 minutes as Host 1 (correct), 30 minutes as Host 2 (correct), and confuses 2 minutes (Host 1 → Host 2). DER = 2 / 60 = 3.3%. Excellent for a clean podcast.
Related terms
Try multi-speaker transcription
Get started — freeLast updated: September 20, 2026
Frequently Asked Questions
What does DER (Diarization Error Rate) mean in transcription?
The standard accuracy metric for speaker diarization: percent of audio time where the system attributes the wrong speaker (or misses speech). State-of-the-art systems achieve 5-15% DER on conversational benchmarks.
Why does DER (Diarization Error Rate) matter when choosing a transcription workflow?
DER (Diarization Error Rate) affects how you understand transcript quality, timing, compatibility, or the technology behind speech-to-text results.
Where can I apply DER (Diarization Error Rate) on Transcript.you?
Use the related workflow at /interview-transcription when you want to see how this glossary concept connects to an actual transcription task.
Browse the full transcription glossary or read the complete guide to AI transcription.