PERIT-STT · Public services

Benefits, permits and civic helplines, where callers read reference numbers aloud, badly.

Audio comes from municipal helplines, benefits intake, permit desks. Scored on word error rate, with a working public services operator as the control.

820
Clips in the target held-out set
63h
Audio, never published
9
Languages in the set
1
Working public services operator as control

Leaderboard

Illustrative figures. Public runs begin when the held-out sets close.
ModelWord error rate (%)Entity accuracyDiarization error
Human control · sector transcriptionistHuman control
2.8 ± 2.3
99%3.3%
Frontier ASR · lab AFrontier
7.4 ± 2.2
90.1%9.5%
Specialist vendor ASRSpecialist
9.0 ± 0.7
86.7%7.8%
Frontier ASR · lab BFrontier
11.1 ± 1.6
89.9%11.5%
Frontier multimodal · lab CFrontier
13.2 ± 2.3
82.9%13.9%
Open weights · 8BOpen weights
14.6 ± 2.2
80%18.8%
Task families

What the set actually asks for.

  • Verbatim transcription
  • Entity capture (IDs, amounts, names)
  • Speaker diarization
  • Code-switching
  • Noisy & far-field audio
  • Domain vocabulary
Method & limits

How a run on this set is produced.

  1. 01Audio is held out and never published. A 60-clip development split per sector is open for calibration.
  2. 02Reference transcripts are written by two annotators independently; a senior reviewer adjudicates every disagreement.
  3. 03Each clip is run 3 times per model. Reported intervals are bootstrapped at 95%.
  4. 04Human control is a working transcriptionist from the same sector, timed and paid at market rate.

Limits, stated up front: 9 languages only, single-channel audio, and no cross-sector transfer. The control is one operator per sector, so the human line carries its own error bar — it is a working standard, not a ceiling.