README.md

metadata

dataset_info:
  config_name: nan_tw
  features:
    - name: id
      dtype: string
    - name: session/stem
      dtype: string
    - name: audio
      dtype: audio
    - name: duration
      dtype: float64
    - name: hanzi
      dtype: string
    - name: lang_code
      dtype: string
    - name: lang_name
      dtype: string
    - name: channel
      dtype: string
    - name: speaker
      dtype: string
    - name: accent
      dtype: string
    - name: has_romaji
      dtype: bool
  splits:
    - name: train
      num_bytes: 8968596247
      num_examples: 74287
    - name: test
      num_bytes: 1478932627
      num_examples: 12551
    - name: eval
      num_bytes: 1335850067
      num_examples: 10619
  download_size: 11845384019
  dataset_size: 11783378941
configs:
  - config_name: nan_tw
    data_files:
      - split: train
        path: nan_tw/train-*
      - split: test
        path: nan_tw/test-*
      - split: eval
        path: nan_tw/eval-*

TEST

Subset	lang_name	hours	n_utts	n_chars_in_utts	secs/utt	chars/sec	n_sents	n_chars_in_sents
nan_tw	Taigi	12.83	12,551	150,718	3.68	3.26	0	0
Total	-	12.83	12,551	150,718	3.68	3.26	0	0

EVAL

Subset	lang_name	hours	n_utts	n_chars_in_utts	secs/utt	chars/sec	n_sents	n_chars_in_sents
nan_tw	Taigi	11.55	10,619	132,245	3.92	3.18	0	0
Total	-	11.55	10,619	132,245	3.92	3.18	0	0

TRAIN

Subset	lang_name	hours	n_utts	n_chars_in_utts	secs/utt	chars/sec	n_sents	n_chars_in_sents
nan_tw	Taigi	78.33	74,287	908,418	3.80	3.22	0	0
Total	-	78.33	74,287	908,418	3.80	3.22	0	0