README.md

metadata

dataset_info:
  - config_name: hak_dp
    features:
      - name: id
        dtype: string
      - name: audio_id
        dtype: string
      - name: audio
        dtype: audio
      - name: duration
        dtype: float64
      - name: hanzi
        dtype: string
      - name: pinyin
        dtype: string
      - name: ipa
        dtype: string
      - name: mandarin
        dtype: string
      - name: lang_code
        dtype: string
      - name: lang_name
        dtype: string
      - name: speaker
        dtype: string
      - name: fsr_kind
        dtype: string
      - name: corpus_kind
        dtype: string
    splits:
      - name: train
        num_bytes: 3064251898
        num_examples: 15888
    download_size: 3146101372
    dataset_size: 3064251898
  - config_name: hak_sx
    features:
      - name: id
        dtype: string
      - name: audio_id
        dtype: string
      - name: audio
        dtype: audio
      - name: duration
        dtype: float64
      - name: hanzi
        dtype: string
      - name: pinyin
        dtype: 'null'
      - name: ipa
        dtype: 'null'
      - name: mandarin
        dtype: 'null'
      - name: lang_code
        dtype: string
      - name: lang_name
        dtype: string
      - name: speaker
        dtype: 'null'
      - name: fsr_kind
        dtype: string
      - name: corpus_kind
        dtype: string
    splits:
      - name: train
        num_bytes: 1660305525
        num_examples: 5868
    download_size: 1633225496
    dataset_size: 1660305525
  - config_name: hak_za
    features:
      - name: id
        dtype: string
      - name: audio_id
        dtype: string
      - name: audio
        dtype: audio
      - name: duration
        dtype: float64
      - name: hanzi
        dtype: string
      - name: pinyin
        dtype: string
      - name: ipa
        dtype: string
      - name: mandarin
        dtype: string
      - name: lang_code
        dtype: string
      - name: lang_name
        dtype: string
      - name: speaker
        dtype: string
      - name: fsr_kind
        dtype: string
      - name: corpus_kind
        dtype: string
    splits:
      - name: train
        num_bytes: 3298371639
        num_examples: 20314
    download_size: 3232042232
    dataset_size: 3298371639
configs:
  - config_name: hak_dp
    data_files:
      - split: train
        path: hak_dp/train-*
  - config_name: hak_sx
    data_files:
      - split: train
        path: hak_sx/train-*
  - config_name: hak_za
    data_files:
      - split: train
        path: hak_za/train-*

TRAIN

Subset	lang_name	hours	n_utts	n_chars_in_utts	secs/utt	chars/sec	n_sents	n_chars_in_sents
hak_dp	Hakka_Dapu	27.27	15,888	271,152	6.18	2.76	0	0
hak_sx	Hakka_Sixian	14.17	5,868	192,713	8.69	3.78	0	0
hak_za	Hakka_Zhaoan	28.01	20,314	312,576	4.96	3.10	0	0
Total	-	69.45	42,070	776,441	5.94	3.11	0	0