✳ Source aggregated · Sourced (per-row source license) · dataset-v2.1

Transcription quality per provider

WER per (provider, language, audio quality).

Origin: Source aggregatedLicense: Sourced (per-row source license)Refresh: quarterlySample size: 5 providersGenerated at: 2026-09-19

Methodology

Per-provider published benchmarks.

Sources

No external sources referenced.

Related tools

Related datasets

Lead path

Inspect my file against this dataset: /tools/validate

Generate correct output via preset: /tools/subtitles/subtitle-to-transcript

Create all target versions in the Creator: /tools/pack