Upload vivoice34.py with huggingface_hub

vivoice34.py

@@ -0,0 +1,74 @@
+import csv
+import os
+
+import datasets
+
+_DESCRIPTION = """
+ViVoice-34 preview dataset with Vietnamese speech samples.
+""".strip()
+
+_CITATION = """
+@dataset{vivoice34_2026,
+  title={ViVoice-34: Vietnamese Speech Dataset from 34 Provinces},
+  year={2026},
+  url={https://huggingface.co/datasets/anonymous-vivoice34/ViVoice34}
+}
+""".strip()
+
+_DATA_DIR = "data/train"
+
+
+class ViVoice34(datasets.GeneratorBasedBuilder):
+    VERSION = datasets.Version("1.0.0")
+
+    def _info(self):
+        features = datasets.Features(
+            {
+                "audio": datasets.Audio(sampling_rate=16000),
+                "file_name": datasets.Value("string"),
+                "transcript": datasets.Value("string"),
+                "speaker_id": datasets.Value("string"),
+                "province": datasets.Value("string"),
+                "gender": datasets.Value("string"),
+                "age_group": datasets.Value("string"),
+                "duration_s": datasets.Value("float32"),
+                "total_word": datasets.Value("int32"),
+                "province_code": datasets.Value("string"),
+            }
+        )
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            citation=_CITATION,
+            features=features,
+        )
+
+    def _split_generators(self, dl_manager):
+        train_dir = dl_manager.download_and_extract(_DATA_DIR)
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                gen_kwargs={"data_dir": train_dir},
+            )
+        ]
+
+    def _generate_examples(self, data_dir):
+        metadata_path = os.path.join(data_dir, "metadata.csv")
+        with open(metadata_path, "r", encoding="utf-8", newline="") as handle:
+            reader = csv.DictReader(handle)
+            for idx, row in enumerate(reader):
+                file_name = row.get("file_name", "")
+                audio_path = os.path.join(data_dir, file_name)
+                duration = float(row.get("duration_s") or 0.0)
+                total_word = int(row.get("total_word") or 0)
+                yield idx, {
+                    "audio": audio_path,
+                    "file_name": file_name,
+                    "transcript": row.get("transcript", ""),
+                    "speaker_id": row.get("speaker_id", ""),
+                    "province": row.get("province", ""),
+                    "gender": row.get("gender", ""),
+                    "age_group": row.get("age_group", ""),
+                    "duration_s": duration,
+                    "total_word": total_word,
+                    "province_code": row.get("province_code", ""),
+                }