Reduce unit test execution time (by 50%)

open_dubbing/text_to_speech.py

@@ -235,7 +235,7 @@ def get_start_time_of_next_speech_utterance(
         start: float,
         end: float,
         audio_file: str,
-    ) -> int:
+    ) -> float:
         result = None
         for utterance in utterance_metadata:
             u_start = utterance["start"]

tests/speech_to_text_faster_whisper_test.py

@@ -18,26 +18,27 @@
 
 
 class TestTextToSpeechFasterWhisper:
+    @classmethod
+    def setup_class(cls):
+        """Set up the class by loading the model once."""
+        cls.stt = SpeechToTextFasterWhisper()
+        cls.stt.load_model()
 
     def test_transcribe(self):
         data_dir = os.path.dirname(os.path.realpath(__file__))
         filename = os.path.join(data_dir, "data/this_is_a_test.mp3")
-        stt = SpeechToTextFasterWhisper()
-        stt.load_model()
-        text = stt._transcribe(vocals_filepath=filename, source_language_iso_639_1="en")
+        text = self.stt._transcribe(
+            vocals_filepath=filename, source_language_iso_639_1="en"
+        )
         assert text.strip() == "This is a test."
 
     def test_detect_language(self):
         data_dir = os.path.dirname(os.path.realpath(__file__))
         filename = os.path.join(data_dir, "data/this_is_a_test.mp3")
-        stt = SpeechToTextFasterWhisper()
-        stt.load_model()
-        language = stt.detect_language(filename)
+        language = self.stt.detect_language(filename)
         assert language == "ell"
 
     def test_get_languages(self):
-        stt = SpeechToTextFasterWhisper()
-        stt.load_model()
-        languages = stt.get_languages()
+        languages = self.stt.get_languages()
         assert len(languages) == 100
         assert "eng" in languages

tests/speech_to_text_whisper_transformers_test.py

@@ -21,26 +21,26 @@
 
 class TestSpeechToTextWhisperTransformers:
 
+    @classmethod
+    def setup_class(cls):
+        cls.data_dir = os.path.dirname(os.path.realpath(__file__))
+        cls.stt = SpeechToTextWhisperTransformers()
+        cls.stt.load_model()
+
     def test_transcribe(self):
-        data_dir = os.path.dirname(os.path.realpath(__file__))
-        filename = os.path.join(data_dir, "data/this_is_a_test.mp3")
-        stt = SpeechToTextWhisperTransformers()
-        stt.load_model()
-        text = stt._transcribe(vocals_filepath=filename, source_language_iso_639_1="en")
+        filename = os.path.join(self.data_dir, "data/this_is_a_test.mp3")
+        text = self.stt._transcribe(
+            vocals_filepath=filename, source_language_iso_639_1="en"
+        )
         assert text.strip() == "This is a test."
 
     def test_detect_language(self):
-        data_dir = os.path.dirname(os.path.realpath(__file__))
-        filename = os.path.join(data_dir, "data/this_is_a_test.mp3")
-        stt = SpeechToTextWhisperTransformers()
-        stt.load_model()
-        language = stt.detect_language(filename)
+        filename = os.path.join(self.data_dir, "data/this_is_a_test.mp3")
+        language = self.stt.detect_language(filename)
         # Returns 'ell' instead of 'eng'. We check the size only
         assert len(language) == 3
 
     def test_get_languages(self):
-        stt = SpeechToTextWhisperTransformers()
-        stt.load_model()
-        languages = stt.get_languages()
+        languages = self.stt.get_languages()
         assert len(languages) == 100
         assert "eng" in languages

tests/text_to_speech_test.py

@@ -233,11 +233,9 @@ def test_dub_utterances_modified_one_modification_decrease_speed(self):
         # Mock dependencies
         utterance_metadata = self._get_dub_metadata()
         utterance_metadata[1]["speed"] = "1.2"
-        print(f"utterance_metadata: {utterance_metadata}")
 
         update_metadata = self._get_dub_metadata()
         del update_metadata[0]
-        print(f"update_metadata: {update_metadata}")
 
         output_directory = "/output"
         target_language = "eng"

tests/utterance_test.py

@@ -20,7 +20,7 @@
 from open_dubbing.utterance import Utterance
 
 
-class TestUterrance:
+class TestUtterance:
 
     def testrun_save_utterance(self):
         with tempfile.TemporaryDirectory() as temp_dir:

tests/voice_gender_classifier_test.py

@@ -18,12 +18,14 @@
 import pytest
 import torch
 
-from open_dubbing.voice_gender_classifier import (  # Assuming this is the file name
-    VoiceGenderClassifier,
-)
+from open_dubbing.voice_gender_classifier import VoiceGenderClassifier
 
 
 class TestVoiceGenderClassifier:
+    @classmethod
+    def setup_class(cls):
+        """Set up the VoiceGenderClassifier model once for all tests."""
+        cls.classifier = VoiceGenderClassifier()
 
     @pytest.mark.parametrize(
         "logits_gender, expected_gender",
@@ -33,17 +35,14 @@ class TestVoiceGenderClassifier:
         ],
     )
     def test_interpret_gender(self, logits_gender, expected_gender):
-        classifier = VoiceGenderClassifier()
-        predicted_gender = classifier._interpret_gender(logits_gender)
-
+        predicted_gender = self.classifier._interpret_gender(logits_gender)
         assert predicted_gender == expected_gender
 
     def test_load_audio_file(self):
         data_dir = os.path.dirname(os.path.realpath(__file__))
         filename = os.path.join(data_dir, "data/this_is_a_test.mp3")
 
-        classifier = VoiceGenderClassifier()
-        samples, target_sampling_rate = classifier.load_audio_file(filename)
+        samples, target_sampling_rate = self.classifier.load_audio_file(filename)
         sample_sum = np.sum(samples)
         assert 16000 == target_sampling_rate
         assert np.isclose(sample_sum, -19.797165, atol=2)
@@ -52,6 +51,5 @@ def test_get_gender_for_file(self):
         data_dir = os.path.dirname(os.path.realpath(__file__))
         filename = os.path.join(data_dir, "data/this_is_a_test.mp3")
 
-        classifier = VoiceGenderClassifier()
-        gender = classifier.get_gender_for_file(filename)
+        gender = self.classifier.get_gender_for_file(filename)
         assert "Male" == gender