Fact · Dictation software
For the Open ASR Leaderboard's normalized transcripts, punctuation and casing are removed, along with disfluencies such as fillers, repetitions and repairs, before word error rate is computed.
Checked against its source on October 16, 2026
In the source's own words
“For normalized transcriptions, punctuation and casing is removed, as well as disfluencies such as fillers”
Open ASR Leaderboard: Towards Reproducible and Transparent Multilingual and Long-Form Speech Recognition Evaluation, Hugging Face and collaborators, arXiv, 2025, Evaluation methodology, normalized transcriptions.
Open the original- What it measures
- The paper says the normalized scoring applies to its short-form evaluation, and that some datasets keep punctuation or disfluencies in their own reference transcripts.
- What it does not show
- The scoring rule makes the leaderboard numbers easier to compare, but it means they do not measure punctuation or the um and uh a writer may want removed. A dictation test should count the errors you care about.
Source
Cited on
- Speech Recognition: What It Is and How Well It WorksTopic
- Meeting Transcription: How It Works and What to UseGuide
- Speech-to-Text APIs Compared: Price, Accuracy, SpeedGuide
- Word Error Rate (WER): How to Calculate and Read ItGuide
- How Accurate Is Speech Recognition? The Real NumbersGuide
- Speech to Text and LLMs: What Language Models ChangeGuide
- Speech Recognition Models: How Neural ASR WorksGuide
- Human vs AI Transcription: Accuracy, Cost, SpeedGuide
- Open-Source Speech Recognition: The Best ModelsGuide
- Speech Recognition Benchmarks and LeaderboardsGuide
- How do you learn speech recognition?Question
- What is the most accurate dictation software?Question
- What is the Open ASR Leaderboard?Question
- Which speech-to-text API is the best?Question
- Why do speech-to-text accuracy claims disagree?Question
- Filler wordsGlossary
- Word error rate (WER)Glossary