| --- |
| configs: |
| - config_name: hak |
| data_files: |
| - split: train |
| path: corpus/hak.train.clean.jsonl |
| - split: eval |
| path: corpus/hak.eval.clean.jsonl |
| - config_name: hak-5000h |
| data_files: |
| - split: train |
| path: corpus/hak.train.coverage-5000h.jsonl |
| - split: eval |
| path: corpus/hak.eval.clean.jsonl |
| - config_name: audio-drama_wointian |
| data_files: |
| - split: train |
| path: audio/drama_wointian/*/part-*.parquet |
| - config_name: audio-hak_yijiazhizhu |
| data_files: |
| - split: train |
| path: audio/hak_yijiazhizhu/*/part-*.parquet |
| - config_name: audio-hakka_cinema_001 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_001/*/part-*.parquet |
| - config_name: audio-hakka_cinema_002 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_002/*/part-*.parquet |
| - config_name: audio-hakka_cinema_003 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_003/*/part-*.parquet |
| - config_name: audio-hakka_cinema_005 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_005/*/part-*.parquet |
| - config_name: audio-hakka_cinema_007 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_007/*/part-*.parquet |
| - config_name: audio-hakka_cinema_008 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_008/*/part-*.parquet |
| - config_name: audio-hakka_cinema_010 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_010/*/part-*.parquet |
| - config_name: audio-hakka_cinema_011 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_011/*/part-*.parquet |
| - config_name: audio-hakka_cinema_017 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_017/*/part-*.parquet |
| - config_name: audio-hakka_cinema_019 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_019/*/part-*.parquet |
| - config_name: audio-hakka_cinema_022 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_022/*/part-*.parquet |
| - config_name: audio-hakka_cinema_023 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_023/*/part-*.parquet |
| - config_name: audio-hakka_cinema_027 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_027/*/part-*.parquet |
| - config_name: audio-hakka_cinema_029 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_029/*/part-*.parquet |
| - config_name: audio-hakka_cinema_031 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_031/*/part-*.parquet |
| - config_name: audio-hakka_cinema_040 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_040/*/part-*.parquet |
| - config_name: audio-hakka_cinema_045 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_045/*/part-*.parquet |
| - config_name: audio-hakka_cinema_049 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_049/*/part-*.parquet |
| - config_name: audio-hakka_cinema_052 |
| data_files: |
| - split: train |
| path: audio/hakka_cinema_052/*/part-*.parquet |
| - config_name: audio-hakka_processed_2023_year_hakka_collection |
| data_files: |
| - split: train |
| path: audio/hakka_processed_2023_year_hakka_collection/*/part-*.parquet |
| - config_name: audio-hakka_processed_2024_year_hakka_collection |
| data_files: |
| - split: train |
| path: audio/hakka_processed_2024_year_hakka_collection/*/part-*.parquet |
| - config_name: audio-hakka_processed_2025_year_hakka_collection |
| data_files: |
| - split: train |
| path: audio/hakka_processed_2025_year_hakka_collection/*/part-*.parquet |
| - config_name: audio-hakka_processed_2026_year_hakka_collection |
| data_files: |
| - split: train |
| path: audio/hakka_processed_2026_year_hakka_collection/*/part-*.parquet |
| - config_name: audio-hakka_processed_audio_fixed_25146 |
| data_files: |
| - split: train |
| path: audio/hakka_processed_audio_fixed_25146/*/part-*.parquet |
| - config_name: audio-hakka_processed_eattogther |
| data_files: |
| - split: train |
| path: audio/hakka_processed_eattogther/*/part-*.parquet |
| - config_name: audio-hakka_processed_show_hou |
| data_files: |
| - split: train |
| path: audio/hakka_processed_show_hou/*/part-*.parquet |
| - config_name: audio-hakka_processed_singing |
| data_files: |
| - split: train |
| path: audio/hakka_processed_singing/*/part-*.parquet |
| - config_name: audio-hakka_processed_whotoeat |
| data_files: |
| - split: train |
| path: audio/hakka_processed_whotoeat/*/part-*.parquet |
| - config_name: audio-hakka_tts-concat-dapu |
| data_files: |
| - split: train |
| path: audio/hakka_tts/concat/dapu/*/part-*.parquet |
| - config_name: audio-hakka_tts-concat-hailu |
| data_files: |
| - split: train |
| path: audio/hakka_tts/concat/hailu/*/part-*.parquet |
| - config_name: audio-hakka_tts-concat-nansixian |
| data_files: |
| - split: train |
| path: audio/hakka_tts/concat/nansixian/*/part-*.parquet |
| - config_name: audio-hakka_tts-concat-raoping |
| data_files: |
| - split: train |
| path: audio/hakka_tts/concat/raoping/*/part-*.parquet |
| - config_name: audio-hakka_tts-concat-sixian |
| data_files: |
| - split: train |
| path: audio/hakka_tts/concat/sixian/*/part-*.parquet |
| - config_name: audio-hakka_tts-concat-zhaoan |
| data_files: |
| - split: train |
| path: audio/hakka_tts/concat/zhaoan/*/part-*.parquet |
| - config_name: audio-hakka_tts-elearning-dapu |
| data_files: |
| - split: train |
| path: audio/hakka_tts/elearning/dapu/*/part-*.parquet |
| - config_name: audio-hakka_tts-elearning-hailu |
| data_files: |
| - split: train |
| path: audio/hakka_tts/elearning/hailu/*/part-*.parquet |
| - config_name: audio-hakka_tts-elearning-raoping |
| data_files: |
| - split: train |
| path: audio/hakka_tts/elearning/raoping/*/part-*.parquet |
| - config_name: audio-hakka_tts-elearning-sixian |
| data_files: |
| - split: train |
| path: audio/hakka_tts/elearning/sixian/*/part-*.parquet |
| - config_name: audio-hakka_tts-elearning-zhaoan |
| data_files: |
| - split: train |
| path: audio/hakka_tts/elearning/zhaoan/*/part-*.parquet |
| - config_name: audio-hakka_tts-moe-dapu |
| data_files: |
| - split: train |
| path: audio/hakka_tts/moe/dapu/*/part-*.parquet |
| - config_name: audio-hakka_tts-moe-hailu |
| data_files: |
| - split: train |
| path: audio/hakka_tts/moe/hailu/*/part-*.parquet |
| - config_name: audio-hakka_tts-moe-nansixian |
| data_files: |
| - split: train |
| path: audio/hakka_tts/moe/nansixian/*/part-*.parquet |
| - config_name: audio-hakka_tts-moe-raoping |
| data_files: |
| - split: train |
| path: audio/hakka_tts/moe/raoping/*/part-*.parquet |
| - config_name: audio-hakka_tts-moe-sixian |
| data_files: |
| - split: train |
| path: audio/hakka_tts/moe/sixian/*/part-*.parquet |
| - config_name: audio-hakka_tts-moe-zhaoan |
| data_files: |
| - split: train |
| path: audio/hakka_tts/moe/zhaoan/*/part-*.parquet |
| - config_name: audio-上家下屋 |
| data_files: |
| - split: train |
| path: audio/上家下屋/*/part-*.parquet |
| - config_name: audio-客家戲曲 |
| data_files: |
| - split: train |
| path: audio/客家戲曲/*/part-*.parquet |
| - config_name: audio-客庄好味道 |
| data_files: |
| - split: train |
| path: audio/客庄好味道/*/part-*.parquet |
| - config_name: audio-巷弄裡的吉光片羽 |
| data_files: |
| - split: train |
| path: audio/巷弄裡的吉光片羽/*/part-*.parquet |
| - config_name: audio-暗香風華 |
| data_files: |
| - split: train |
| path: audio/暗香風華/*/part-*.parquet |
| - config_name: audio-溜麵線雜貨店 |
| data_files: |
| - split: train |
| path: audio/溜麵線雜貨店/*/part-*.parquet |
| - config_name: audio-維基百客 |
| data_files: |
| - split: train |
| path: audio/維基百客/*/part-*.parquet |
| --- |
| |
| # hak cleaned speech corpus |
|
|
| This refresh provides a converged full train manifest, a frozen eval manifest, |
| and a token-coverage subset of approximately 5000 hours. All three manifests |
| reuse the existing validated FLAC-in-Parquet audio pool; no audio was copied, |
| repacked, or transcoded for this refresh. |
|
|
| ## Corpus overview |
|
|
| | Variant | Rows | Hours | |
| |---|---:|---:| |
| | train | 6,638,601 | 16435.239 | |
| | coverage_5000h | 1,668,639 | 4999.993 | |
| | eval | 823 | 2.001 | |
| |
| All audio locators are portable and have the form |
| `audio/<subset>/<group>/part-N.parquet#row=N`. |
| |
| The release audit records exact input/output hashes, rejected audio mappings, |
| language-tag validation, and zero train/eval audio-identity overlap. |
| |
| ```python |
| from datasets import load_dataset |
| ds = load_dataset("WTFO/hak", "hak-5000h", split="train") |
| ``` |
| |
