Datasets:
metadata
license: other
task_categories:
- automatic-speech-recognition
- text-to-speech
language:
- id
YuKuanFu/id
Corpus JSONL files are in corpus_dir_jsonl/.
Source-set corpus, manifests, and audio payload shards are in subsets/<source_set>/.
Only audio referenced by JSONL rows is included; source audio directories are never uploaded wholesale.
WAV source audio is converted to FLAC before Parquet writing.
corpus_dir_jsonl/{train,train_stage1,eval}.jsonl
subsets/<source_set>/
corpus_dir_jsonl/{train,train_stage1,eval}.jsonl
manifests/audio_manifest.tsv
audio_parquet/part-xxxxx.parquet