sync ms

README.md

@@ -4,24 +4,18 @@ viewer: false
 ---
 
 # Genshin Impact & Honkai Star Rail Game Character Voice Dataset
-This repository is the integration code for the [aihobbyist/Genshin_Dataset](https://www.modelscope.cn/datasets/aihobbyist/Genshin_Dataset) provided by [AI Hobbyist](https://www.modelscope.cn/organization/aihobbyist), with the final right of interpretation belonging to [MiHoYo](https://www.mihoyo.com). The integration code aims to provide a more convenient usage solution for the community: for `Python` developers, only a few lines of code are needed to automatically search, download, split by language, and normalize on demand, instead of manually searching for and downloading the cumbersome complete package to accommodate users with different backgrounds and thresholds. The integration code is provided by [Genius-Society](https://www.modelscope.cn/organization/Genius-Society). If you need to use the compressed package version of the data source and write your own data processing scripts, you can go to [aihobbyist/Genshin_Dataset](https://www.modelscope.cn/datasets/aihobbyist/Genshin_Dataset) to support the original author.
-
-## Environment
-```bash
-pip install datasets py7zr librosa
-```
+This repository is the integration code for the TTS datasets of miHoYo games provided by [AI Hobbyist](https://www.modelscope.cn/organization/aihobbyist), with the final right of interpretation belonging to [miHoYo](https://www.mihoyo.com). The integration code aims to provide a more convenient usage solution for the community: for `Python` developers, only a few lines of code are needed to automatically search, download, split by language, and normalize on demand, instead of manually searching for and downloading the cumbersome complete package to accommodate users with different backgrounds and thresholds. The integration code is provided by [Genius-Society](https://www.modelscope.cn/organization/Genius-Society). If you need to use the compressed package version of the data source and write your own data processing scripts, you can support the original author.
 
 ## Supported characters
-[Genshin characters](https://pan.ai-hobbyist.com/Genshin%20Datasets/%E8%8B%B1%E8%AF%AD)<br>
-[Star Rail characters](https://pan.ai-hobbyist.com/StarRail%20Datasets/%E5%88%86%E8%A7%92%E8%89%B2%E5%8C%85/%E8%8B%B1%E8%AF%AD)
+<https://res.acgnai.top>
 
 ## Splits
 | Language | Genshin Impact | Honkai Star Rail |
 | :------: | :------------: | :--------------: |
-| Chinese  |  `genshin_zh`  |   `starail_zh`   |
-| Japanese |  `genshin_jp`  |   `starail_jp`   |
-| English  |  `genshin_en`  |   `starail_en`   |
-|  Korean  |  `genshin_kr`  |   `starail_kr`   |
+| Chinese  |  `Genshin_zh`  |  `StarRail_zh`   |
+| Japanese |  `Genshin_ja`  |  `StarRail_ja`   |
+| English  |  `Genshin_en`  |  `StarRail_en`   |
+|  Korean  |  `Genshin_ko`  |  `StarRail_ko`   |
 
 ## Usage
 ```python
@@ -30,7 +24,7 @@ from datasets import load_dataset
 ds = load_dataset(
     "Genius-Society/hoyoTTS",
     name="黑塔",  # The Chinese name of the character
-    split="starail_en",  # The language options should match the character's game
+    split="StarRail_en",  # The language options should match the character's game
     cache_dir="./__pycache__",  # The cache location of the dataset
 )
 for item in ds:
@@ -46,7 +40,7 @@ cd hoyoTTS
 ## Mirror
 <https://www.modelscope.cn/datasets/Genius-Society/hoyoTTS>
 
-## References
-[1] <https://pan.ai-hobbyist.com><br>
-[2] <https://ys.mihoyo.com/main><br>
-[3] <https://sr.mihoyo.com>
+## Thanks
+ - <https://pan.ai-hobbyist.com>
+ - <https://ys.mihoyo.com/main>
+ - <https://sr.mihoyo.com>

hoyoTTS.py

@@ -1,20 +1,13 @@
 import os
 import random
 import datasets
-import subprocess
+import requests
 
+_URL = "https://res.ai-lab.top/api/acgnailib/models"
 
-_DOMAIN = "https://pan.ai-hobbyist.com/d"
-_URLS = {
-    "genshin_zh": "Genshin Datasets/中文",
-    "genshin_jp": "Genshin Datasets/日语",
-    "genshin_en": "Genshin Datasets/英语",
-    "genshin_kr": "Genshin Datasets/韩语",
-    "starail_zh": "StarRail Datasets/分角色包/中文",
-    "starail_jp": "StarRail Datasets/分角色包/日语",
-    "starail_en": "StarRail Datasets/分角色包/英语",
-    "starail_kr": "StarRail Datasets/分角色包/韩语",
-}
+_GAMES = {"Genshin": "原神", "StarRail": "星穹铁道"}
+
+_LANGUAGES = {"zh": "中文", "ja": "日语", "en": "英语", "ko": "韩语"}
 
 
 class hoyoTTS(datasets.GeneratorBasedBuilder):
@@ -42,33 +35,55 @@ class hoyoTTS(datasets.GeneratorBasedBuilder):
 
         return content.strip()
 
+    def _parse_url(self, game: str, lang: str):
+        response = requests.post(
+            _URL,
+            json={
+                "category": _GAMES[game],
+                "repo": f"datasets/aihobbyist/{game}_Dataset",
+                "root_path": "/datasets",
+                "subcategory": _LANGUAGES[lang],
+            },
+            headers={
+                "user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/141.0.0.0 Safari/537.36 Edg/141.0.0.0"
+            },
+        )
+        response.raise_for_status()
+        data = response.json()
+        for item in data:
+            if item["dataname"] == self.config.name:
+                return item["dl_link"]
+
+        return None
+
     def _split_generators(self, dl_manager):
-        datasplits = []
-        subprocess.run(["pip", "install", "py7zr", "librosa"])
-        for region in _URLS:
-            url = f"{_DOMAIN}/{_URLS[region]}/{self.config.name}.7z"
-            try:
-                data_files = dl_manager.download_and_extract(url)
-
-            except Exception as e:
-                print(f"{e}, retrying...")
-                data_files = dl_manager.download_and_extract(url)
-
-            if os.path.isdir(data_files):
+        data_splits = []
+        for game in _GAMES:
+            for lang in _LANGUAGES:
+                url = self._parse_url(game, lang)
+                if not url:
+                    continue
+
+                try:
+                    data_files = dl_manager.download_and_extract(url)
+                except Exception as e:
+                    print(f"{e}, retrying...")
+                    data_files = dl_manager.download_and_extract(url)
+
                 files = []
                 for fpath in dl_manager.iter_files([data_files]):
                     if os.path.basename(fpath).endswith(".wav"):
                         files.append({"audio": fpath, "text": self._get_txt(fpath)})
 
                 random.shuffle(files)
-                datasplits.append(
+                data_splits.append(
                     datasets.SplitGenerator(
-                        name=region,
+                        name=f"{game}_{lang}",
                         gen_kwargs={"files": files},
                     )
                 )
 
-        return datasplits
+        return data_splits
 
     def _generate_examples(self, files):
         for i, path in enumerate(files):