STT-DATASET/en-accent-edacc

edacc

ASR-ready version of edinburghcstr/edacc (license: CC-BY-SA-4.0).

Edinburgh International Accents of English: conversational Zoom speech, ~40 L1/accents. Source validation -> train here, source test kept as test (use it as an eval set).

Format: parquet, 500-row row groups (Cohere training format). Columns: audio (bytes, 16 kHz mono FLAC), id, text, duration, accent, speaker_id, gender, source, lang, reference_text, meta (JSON string). Clips: 0.5–30 s, non-empty text. No judge/quality filtering applied.

Splits: test: 6,910 clips / 11.5 h, train: 7,388 clips / 11.9 h

Top accent labels: Southern British English 1,013, Nigerian English 992, Irish English 962, Kenyan English 892, Spanish 847, Romanian 798, Vietnamese 787, Catalan 607, Jamaican English 603, Mandarin 549, Lithuanian 527, Indian English 523, Mainstream US English 497, Italian 480, Polish 396