Refactor CMUARCTIC unittest (#1147)

Co-authored-by: krishnakalyan3 <skalyan@cloudera.com>

test/torchaudio_unittest/datasets/cmuarctic_test.py

@@ -12,19 +12,15 @@ from torchaudio_unittest.common_utils import (
 )
 
 
-class TestCMUARCTIC(TempDirMixin, TorchaudioTestCase):
-    backend = "default"
-
-    root_dir = None
-    samples = []
-
-    @classmethod
-    def setUpClass(cls):
-        cls.root_dir = cls.get_base_temp_dir()
+def get_mock_dataset(root_dir):
+    """
+    root_dir: directory to the mocked dataset
+    """
+    mocked_data = []
     sample_rate = 16000
     utterance = "This is a test utterance."
 
-        base_dir = os.path.join(cls.root_dir, "ARCTIC", "cmu_us_aew_arctic")
+    base_dir = os.path.join(root_dir, "ARCTIC", "cmu_us_aew_arctic")
     txt_dir = os.path.join(base_dir, "etc")
     os.makedirs(txt_dir, exist_ok=True)
     txt_file = os.path.join(txt_dir, "txt.done.data")
@@ -51,9 +47,22 @@ class TestCMUARCTIC(TempDirMixin, TorchaudioTestCase):
                     utterance,
                     utterance_id.split("_")[1],
                 )
-                    cls.samples.append(sample)
+                mocked_data.append(sample)
                 txt.write(f'( {utterance_id} "{utterance}" )\n')
                 seed += 1
+    return mocked_data
+
+
+class TestCMUARCTIC(TempDirMixin, TorchaudioTestCase):
+    backend = "default"
+
+    root_dir = None
+    samples = []
+
+    @classmethod
+    def setUpClass(cls):
+        cls.root_dir = cls.get_base_temp_dir()
+        cls.samples = get_mock_dataset(cls.root_dir)
 
     def _test_cmuarctic(self, dataset):
         n_ite = 0