load script

dusha.py

@@ -24,8 +24,8 @@ class Dusha(datasets.GeneratorBasedBuilder):
             description=_DESCRIPTION,
             features=datasets.Features(
                 {
-                    "path": datasets.Value("string"),
-                    "audio": datasets.features.Audio(sampling_rate=16_000),
+                    "file": datasets.Value("string"),
+                    "audio": datasets.Audio(sampling_rate=16_000),
                     "label": datasets.Value("string"),
                 }
             ),
@@ -35,30 +35,32 @@ class Dusha(datasets.GeneratorBasedBuilder):
 
     def _split_generators(self, dl_manager):
         metadata = dl_manager.download(_METADATA_URL)
-        archive = dl_manager.download(_DATA_URL)
+        archive = dl_manager.download_and_extract(_DATA_URL)
         return [datasets.SplitGenerator(
-            name=datasets.Split.TRAIN,
+            name=datasets.Split.ALL,
             gen_kwargs={
-                "audio_files": dl_manager.iter_archive(archive),
+                "audio_files": archive,
                 "metadata": metadata},
         )]
 
     def _generate_examples(self, audio_files, metadata: str):
-        examples = dict()
+        examples = list()
 
         with open(metadata, encoding="utf-8") as f:
             csv_reader = csv.reader(f, delimiter=",", skipinitialspace=True)
             next(csv_reader)
             for row in csv_reader:
                 audio_path, label = row
+                full_audio_path = os.path.join(audio_files, audio_path)
+                with open(full_audio_path) as audio_file:
+                    audio = {"path": full_audio_path, "bytes": audio_file.read()}
                 res = dict()
-                res["path"] = os.path.join(audio_files, audio_path)
+                res["file"] = audio_path.split("/")[1]
+                res["audio"] = audio
                 res["label"] = label
-                examples[audio_path] = res
+                examples.append(res)
 
         key = 0
-        for path, f in audio_files:
-            if path in examples:
-                audio = {"path": path, "bytes": f.read()}
-                yield key, {**examples[path], "audio": audio}
-                key += 1
+        for example in examples:
+            yield key, example
+            key += 1