Update wmms.py

wmms.py

@@ -4,11 +4,13 @@
 
 
 import os
+import random
 import textwrap
 import datasets
 import itertools
 import typing as tp
 from pathlib import Path
+from collections import defaultdict
 from sklearn.model_selection import train_test_split
 
 SAMPLE_RATE = 16_000
@@ -57,18 +59,40 @@ class WMMS(datasets.GeneratorBasedBuilder):
         """Returns SplitGenerators."""
         archive_path = dl_manager.extract(_COMPRESSED_FILENAME)
         extensions = ['.wav']
-        _, _walker = fast_scandir(archive_path, extensions, recursive=True)
+        _, filepaths = fast_scandir(archive_path, extensions, recursive=True)
+        labels = [default_find_classes(f) for f in filepaths]
 
-        train_walker, test_walker = train_test_split(
-            _walker, test_size=0.2, random_state=914, stratify=[default_find_classes(f) for f in _walker]
-        )
+        # train_walker, test_walker = train_test_split(
+        #     _walker, test_size=0.2, random_state=914, stratify=[default_find_classes(f) for f in _walker]
+        # )
+
+        # Step 1: Organize samples by class
+        class_to_files = defaultdict(list)
+        for filepath, label in zip(filepaths, labels):
+            class_to_files[label].append(filepath)
+        
+        # Step 2: Select exactly 2 samples per class for the test set
+        test_files, test_labels = [], []
+        train_files, train_labels = [], []
+        
+        for label, files in class_to_files.items():
+            if len(files) < 2:
+                raise ValueError(f"Not enough samples for class {label}")  # Ensure each class has at least 2 samples
+        
+            random.Random(914).shuffle(files)  # Shuffle to ensure randomness
+        
+            test_files.extend(files[:2])  # Pick first 2 for test
+            test_labels.extend([label] * 2)
+        
+            train_files.extend(files[2:])  # Remaining go to train
+            train_labels.extend([label] * (len(files) - 2))
         
         return [
             datasets.SplitGenerator(
-                name=datasets.Split.TRAIN, gen_kwargs={"audio_paths": train_walker, "split": "train"}
+                name=datasets.Split.TRAIN, gen_kwargs={"audio_paths": train_files, "split": "train"}
             ), 
             datasets.SplitGenerator(
-                name=datasets.Split.TEST, gen_kwargs={"audio_paths": test_walker, "split": "test"}
+                name=datasets.Split.TEST, gen_kwargs={"audio_paths": test_files, "split": "test"}
             ), 
         ]