{
 "id": "committa/serena-synthetic-it-27h",
 "owner": "committa",
 "name": "serena-synthetic-it-27h",
 "downloads": 1111,
 "likes": 1,
 "last_modified": "2026-08-03T12:00:50.000Z",
 "stated_licence": "cc-by-4.0",
 "stated_from": "tag",
 "licence_name": null,
 "licence_link": null,
 "file_licence": "cc-by-4.0",
 "licence_files": [
  "LICENSE"
 ],
 "sources": [],
 "gated": false,
 "tasks": [
  "text-to-speech"
 ],
 "citation": "The sentence corpus (29,475 sentences) is a mix of two sources:\n\n- **~90% Tatoeba-derived sentences** (Italian, CC-BY 2.0 FR, from the official export at\n  <https://downloads.tatoeba.org/exports/per_language/ita/ita_sentences.tsv.bz2>),\n  **adapted**: English/American given names were replaced with Italian names, and the text\n  received editorial corrections and accent fixes. They are therefore derivative/adapted\n  works, not verbatim copies.\n- **~10% original LLM-authored sentences** across conversational, technical and literary\n  registers, written specifically for this corpus (models: Claude Sonnet, GPT-4o, Qwen,\n  Gemini, DeepSeek; one production batch was excluded after a quality review).\n\nThe corpus was assembled with deterministic seed=42 shuffles; the name-replacement and\nLLM batch verification reports are included in the\n[qwen3-tts-dataset-generator](https://github.com/Committa/qwen3-tts-dataset-generator) project.",
 "signal": "permissive",
 "reason": "The declared licence (cc-by-4.0) is classified as permissive.",
 "revision": "30544c16e43b4d1414a0cc029ff6cd377f475ff2",
 "stated_evidence": "https://huggingface.co/datasets/committa/serena-synthetic-it-27h/blob/30544c16e43b4d1414a0cc029ff6cd377f475ff2/README.md",
 "file_evidence": "https://huggingface.co/datasets/committa/serena-synthetic-it-27h/blob/30544c16e43b4d1414a0cc029ff6cd377f475ff2/LICENSE",
 "card_nc": false
}