supporting audio zip files

snow-mountain.py

@@ -80,7 +80,7 @@ class Test(datasets.GeneratorBasedBuilder):
         features = datasets.Features(
             {
                 "sentence": datasets.Value("string"),
-                # "audio": datasets.features.Audio(sampling_rate=16_000),
+                "audio": datasets.features.Audio(sampling_rate=16_000),
                 "path": datasets.Value("string"),
             }
         )
@@ -137,22 +137,22 @@ class Test(datasets.GeneratorBasedBuilder):
             data_df = pd.read_csv(f,sep=',')
             audio_data = {}
             for index,row in data_df.iterrows():
-                # audio = row['path'].split('/')[-1]
-                # if audio in audio_data:
-                #     content = audio_data[audio]
-                # else:    
-                #     zip_url = '/'.join(row["path"].split('/')[:-1])+'.zip'
-                #     archive_path = dl_manager.download(zip_url)
-                #     for path, file in dl_manager.iter_archive(archive_path):
-                #         audio_ = path.split('/')[-1]
-                #         if audio_ not in audio_data:
-                #             content = file.read()
-                #             audio_data[audio_] = content
-                #     if audio in audio_data:
-                #         content = audio_data[audio]
+                audio = row['path'].split('/')[-1]
+                if audio in audio_data:
+                    content = audio_data[audio]
+                else:    
+                    zip_url = '/'.join(row["path"].split('/')[:-1])+'.zip'
+                    archive_path = dl_manager.download(zip_url)
+                    for path, file in dl_manager.iter_archive(archive_path):
+                        audio_ = path.split('/')[-1]
+                        if audio_ not in audio_data:
+                            content = file.read()
+                            audio_data[audio_] = content
+                    if audio in audio_data:
+                        content = audio_data[audio]
                 yield key, {
                         "sentence": row["sentence"],
                         "path": row["path"],
-                        # "audio":{"path": row["path"], "bytes": content}
+                        "audio":{"path": row["path"], "bytes": content}
                     }
                 key+=1