Update dataset.py

dataset.py

@@ -4,11 +4,11 @@ from datasets import DatasetDict, Dataset, Audio
 from sklearn.model_selection import train_test_split
 
 def load_data(data_dir, test_size=0.2, seed=42):
-    # Load metadata
+
     metadata = pd.read_csv(os.path.join(data_dir, "metadata.csv"))
 
     # Split the data into train and test subsets
-    train_data, test_data = train_test_split(metadata, test_size=test_size, random_state=seed)
+    train_data, test_data = train_test_split(metadata, test_size=0.2, random_state=seed)
 
     # Convert to Hugging Face datasets
     train_dataset = Dataset.from_pandas(train_data)