From ec651457627d5d1fe6b3575a32cf7dccfbd61c8f Mon Sep 17 00:00:00 2001
From: Rob Hallam <0504004h@student.gla.ac.uk>
Date: Sat, 31 Aug 2024 15:08:01 +0100
Subject: [PATCH] test: [WFE] functional test: no audio - no spurious Features

Found out that Whisper throws a hissy fit in the form of a RuntimeError if the
there is no speech in the audio. We should consider catching this.

> RuntimeError: stack expects a non-empty TensorList
> stdout: "No active speech found in audio"

For the moment we can check that no audio throws an error and leave this as a TODO
---
 test/test_feature_extractors_functional.py | 34 +++++++++++++++++++---
 1 file changed, 30 insertions(+), 4 deletions(-)

diff --git a/test/test_feature_extractors_functional.py b/test/test_feature_extractors_functional.py
index 45b2ab6..7da5cff 100644
--- a/test/test_feature_extractors_functional.py
+++ b/test/test_feature_extractors_functional.py
@@ -190,10 +190,7 @@ class TestWordFEFunctional(FEFunctionalTest):
         SAMPLE_VIDEO = f"{self.SAMPLE_DIR}/sample-manual-audio-harvardsentences-rdh-video-colours.mp4"
         DETECT_WORDS = ["birch", "smooth", "chicken", "depth",
                         "juice", "lemons", "box", "thrown", "beside",
-                        "hogs", "fed"]
-        # DETECT_WORDS = ["birch", "smooth", "glue", "chicken", "depth", "well",
-        #                 "juice", "lemons", "punch", "box", "thrown", "beside",
-        #                 "truck", "hogs", "fed"]
+                        "hogs", "fed"] # missing "truck", "glue", "well", "punch" due to problems
 
         # create mock source with the video
         source = mocks.MockSource(path=SAMPLE_VIDEO)
@@ -228,5 +225,34 @@ class TestWordFEFunctional(FEFunctionalTest):
         # check if the word was feature extracted:
         self.assertGreaterEqual(len(testfe.features), 4)
 
+    def test_audio_word_detection_noaudio_nofeatures(self):
+        """Test audio word detection feature extractor
+           Uses:
+             - sample-manual-audio-blank-video-colours.mp4
+               :: silent video (30s)
+                  -- pass if no features extracted
+        """
+
+        SAMPLE_VIDEO = f"{self.SAMPLE_DIR}/sample-manual-audio-blank-video-colours.mp4"
+        DETECT_WORDS = ["birch", "smooth", "chicken", "depth",
+                        "juice", "lemons", "box", "thrown", "beside",
+                        "hogs", "fed"]
+
+        # create mock source with the video
+        source = mocks.MockSource(path=SAMPLE_VIDEO)
+
+        # create the feature extractor
+        testfe = extractors.WordFeatureExtractor(input_files=[source])
+        testfe.setup(words=DETECT_WORDS)
+
+        # ensure no features extracted from blank audio:
+        # self.assertEqual(len(testfe.features), 0)
+        # Actually, Whisper throws a hissy fit if there's no audio:
+        #  RuntimeError: stack expects a non-empty TensorList
+        #  stdout: "No active speech found in audio"
+        # TODO: consider catching this error in the FE
+        with self.assertRaises(RuntimeError):
+            testfe.run()
+
 if __name__ == "__main__":
     unittest.main()