Datasets:
| 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. | |
| ```text | |
| 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 | |
| ``` | |