Intervues

What is ASR word error rate?

Updated 2026-09-16

ASR word error rate (WER) is the percentage of words incorrectly recognized by an automatic speech recognition system — combining substitutions, deletions, and insertions — compared to a reference transcript, used to benchmark speech-to-text accuracy.

How WER is calculated

Speech recognition researchers measure accuracy with word error rate: WER = (Substitutions + Deletions + Insertions) ÷ Total words in reference transcript. A ten-word sentence with one wrong word substituted scores 10% WER. Insertions count extra words the ASR hallucinated; deletions count words it missed entirely.

Lower WER is better. Published benchmarks on clean American English audiobook speech often show single-digit WER on leading models — but hiring interviews are not audiobooks. Accents, code-switching, background noise, cheap phone microphones, and domain vocabulary ('nursing council registration', 'UPI', 'pick-face zone') push real-world WER higher than lab charts suggest.

Interview engines do not expose WER directly to recruiters, but they feel it as wrong follow-up questions, missed knockout answers, or scores attached to transcript text the candidate never said. ASR quality is a first-class hiring risk, not an engineering footnote.

Why accents matter for hiring fairness

Models trained disproportionately on certain accent distributions perform worse on others — documented in academic and industry evaluations across Indian English, regional language-influenced English, and multilingual code-switching common in India hiring pipelines. A candidate judged on a garbled transcript is evaluated on words they did not speak.

Mitigations include accent-robust models, custom vocabulary for role terms, confidence thresholds that trigger repeat requests ('I didn't catch that — could you say it again?'), human review for low-confidence segments, and offering language options where the product supports them.

Fair hiring reviews should ask vendors for WER or accuracy evaluations on speaker populations matching your candidate pool — not only default US English test sets.

WER vs other ASR metrics

MetricWhat it captures
WERWord-level errors vs reference transcript
CERCharacter-level — used for some languages
Real-time factorSpeed of processing vs audio duration
Confidence scoresModel uncertainty per token — useful for gating
Semantic accuracyMeaning preserved despite minor word errors — harder to automate

Implications for AI interviews

  • High WER on knockout questions (license numbers, shift availability) causes false rejections — validate critical fields with confirmation prompts.
  • Scoring rubrics should tolerate minor transcription noise for open-ended answers while flagging low-confidence spans for human review.
  • Multilingual interviews need per-language WER evaluation — a model strong in Hindi may be weak in Tamil or Marathi.
  • Latency and WER interact — rushing ASR finalisation increases errors; tuning is role- and market-specific.

Frequently asked

What WER is good enough for hiring?

There is no universal threshold — depends on question type and downstream handling. Critical structured fields need near-perfect capture; behavioural answers may tolerate higher WER if meaning stays intact.

Does WER measure accent bias directly?

WER differences across demographic speaker groups indicate disparate technical performance — related to fairness risk, though legal analysis requires broader context.

Can humans fix ASR errors after the interview?

Review and correction workflows help for high-stakes decisions but do not scale to ten-thousand-candidate bulk screens — prevention at capture time matters more.

Is WER the same as transcription quality score shown to users?

Products may show qualitative confidence or 'audio quality' warnings rather than raw WER — ask vendors what underlying metrics drive those indicators.

Ready to practise?

Head back to Hiring glossary or start now.

· 3 free credits · pay per interview · nothing recurring

Start practising