load script

dusha.py

@@ -1,5 +1,5 @@
+import csv
 import os
-from typing import Union
 
 import datasets
 
@@ -25,7 +25,7 @@ class Dusha(datasets.GeneratorBasedBuilder):
             features=datasets.Features(
                 {
                     "path": datasets.Value("string"),
-                    "audio": datasets.Audio(sampling_rate=16_000),
+                    "audio": datasets.features.Audio(sampling_rate=16_000),
                     "label": datasets.Value("string"),
                 }
             ),
@@ -37,20 +37,20 @@ class Dusha(datasets.GeneratorBasedBuilder):
         metadata = dl_manager.download(_METADATA_URL)
         archive = dl_manager.download(_DATA_URL)
         return [datasets.SplitGenerator(
-            name=datasets.Split.ALL,
+            name=datasets.Split.TRAIN,
             gen_kwargs={
                 "audio_files": dl_manager.iter_archive(archive),
                 "metadata": metadata},
         )]
 
-    def _generate_examples(self, audio_files: Union[str, os.PathLike], metadata: str):
+    def _generate_examples(self, audio_files, metadata: str):
         examples = dict()
 
         with open(metadata, encoding="utf-8") as f:
-            for row in f:
-                data = row.split(",")
-                audio_path = data[0]
-                label = data[1]
+            csv_reader = csv.reader(f, delimiter=",", skipinitialspace=True)
+            next(csv_reader)
+            for row in csv_reader:
+                audio_path, label = row
                 res = dict()
                 res["path"] = os.path.join(audio_files, audio_path)
                 res["label"] = label