[tests] deflake dither test (#36284)

src/transformers/models/audio_spectrogram_transformer/feature_extraction_audio_spectrogram_transformer.py

@@ -193,7 +193,7 @@ class ASTFeatureExtractor(SequenceFeatureExtractor):
                 )
         else:
             logger.warning(
-                "It is strongly recommended to pass the `sampling_rate` argument to this function. "
+                f"It is strongly recommended to pass the `sampling_rate` argument to `{self.__class__.__name__}()`. "
                 "Failing to do so can result in silent errors that might be hard to debug."
             )
 

src/transformers/models/clap/feature_extraction_clap.py

@@ -308,7 +308,7 @@ class ClapFeatureExtractor(SequenceFeatureExtractor):
                 )
         else:
             logger.warning(
-                "It is strongly recommended to pass the `sampling_rate` argument to this function. "
+                f"It is strongly recommended to pass the `sampling_rate` argument to `{self.__class__.__name__}()`. "
                 "Failing to do so can result in silent errors that might be hard to debug."
             )
 

src/transformers/models/clvp/feature_extraction_clvp.py

@@ -188,7 +188,7 @@ class ClvpFeatureExtractor(SequenceFeatureExtractor):
                 )
         else:
             logger.warning(
-                "It is strongly recommended to pass the `sampling_rate` argument to this function. "
+                f"It is strongly recommended to pass the `sampling_rate` argument to `{self.__class__.__name__}()`. "
                 "Failing to do so can result in silent errors that might be hard to debug."
             )
 

src/transformers/models/dac/feature_extraction_dac.py

@@ -108,7 +108,7 @@ class DacFeatureExtractor(SequenceFeatureExtractor):
                 )
         else:
             logger.warning(
-                "It is strongly recommended to pass the `sampling_rate` argument to this function. "
+                f"It is strongly recommended to pass the `sampling_rate` argument to `{self.__class__.__name__}()`. "
                 "Failing to do so can result in silent errors that might be hard to debug."
             )
 

src/transformers/models/encodec/feature_extraction_encodec.py

@@ -132,7 +132,7 @@ class EncodecFeatureExtractor(SequenceFeatureExtractor):
                 )
         else:
             logger.warning(
-                "It is strongly recommended to pass the `sampling_rate` argument to this function. "
+                f"It is strongly recommended to pass the `sampling_rate` argument to `{self.__class__.__name__}()`. "
                 "Failing to do so can result in silent errors that might be hard to debug."
             )
 

src/transformers/models/musicgen_melody/feature_extraction_musicgen_melody.py

@@ -242,7 +242,7 @@ class MusicgenMelodyFeatureExtractor(SequenceFeatureExtractor):
 
         if sampling_rate is None:
             logger.warning_once(
-                "It is strongly recommended to pass the `sampling_rate` argument to this function. "
+                f"It is strongly recommended to pass the `sampling_rate` argument to `{self.__class__.__name__}()`. "
                 "Failing to do so can result in silent errors that might be hard to debug."
             )
 

src/transformers/models/seamless_m4t/feature_extraction_seamless_m4t.py

@@ -225,7 +225,7 @@ class SeamlessM4TFeatureExtractor(SequenceFeatureExtractor):
                 )
         else:
             logger.warning(
-                "It is strongly recommended to pass the `sampling_rate` argument to this function. "
+                f"It is strongly recommended to pass the `sampling_rate` argument to `{self.__class__.__name__}()`. "
                 "Failing to do so can result in silent errors that might be hard to debug."
             )
 

src/transformers/models/speech_to_text/feature_extraction_speech_to_text.py

@@ -248,7 +248,7 @@ class Speech2TextFeatureExtractor(SequenceFeatureExtractor):
                 )
         else:
             logger.warning(
-                "It is strongly recommended to pass the `sampling_rate` argument to this function. "
+                f"It is strongly recommended to pass the `sampling_rate` argument to `{self.__class__.__name__}()`. "
                 "Failing to do so can result in silent errors that might be hard to debug."
             )
 

src/transformers/models/speecht5/feature_extraction_speecht5.py

@@ -252,7 +252,7 @@ class SpeechT5FeatureExtractor(SequenceFeatureExtractor):
                 )
         else:
             logger.warning(
-                "It is strongly recommended to pass the ``sampling_rate`` argument to this function. "
+                f"It is strongly recommended to pass the `sampling_rate` argument to `{self.__class__.__name__}()`. "
                 "Failing to do so can result in silent errors that might be hard to debug."
             )
 

src/transformers/models/univnet/feature_extraction_univnet.py

@@ -370,7 +370,7 @@ class UnivNetFeatureExtractor(SequenceFeatureExtractor):
                 )
         else:
             logger.warning(
-                "It is strongly recommended to pass the `sampling_rate` argument to this function. "
+                f"It is strongly recommended to pass the `sampling_rate` argument to `{self.__class__.__name__}()`. "
                 "Failing to do so can result in silent errors that might be hard to debug."
             )
 

src/transformers/models/wav2vec2/feature_extraction_wav2vec2.py

@@ -178,7 +178,7 @@ class Wav2Vec2FeatureExtractor(SequenceFeatureExtractor):
                 )
         else:
             logger.warning(
-                "It is strongly recommended to pass the ``sampling_rate`` argument to this function. "
+                f"It is strongly recommended to pass the `sampling_rate` argument to `{self.__class__.__name__}()`. "
                 "Failing to do so can result in silent errors that might be hard to debug."
             )
 

src/transformers/models/whisper/feature_extraction_whisper.py

@@ -255,7 +255,6 @@ class WhisperFeatureExtractor(SequenceFeatureExtractor):
                 Whether or not to return the number of frames of the input raw_speech.
                 These num_frames can be used by the model to compute word level timestamps.
         """
-
         if sampling_rate is not None:
             if sampling_rate != self.sampling_rate:
                 raise ValueError(
@@ -265,7 +264,7 @@ class WhisperFeatureExtractor(SequenceFeatureExtractor):
                 )
         else:
             logger.warning(
-                "It is strongly recommended to pass the `sampling_rate` argument to this function. "
+                f"It is strongly recommended to pass the `sampling_rate` argument to `{self.__class__.__name__}()`. "
                 "Failing to do so can result in silent errors that might be hard to debug."
             )
 

tests/models/speech_to_text/test_feature_extraction_speech_to_text.py

@@ -163,10 +163,10 @@ class Speech2TextFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unitt
 
         # compute features
         input_features_no_dither = feature_extractor_no_dither(
-            np_speech_inputs, padding=True, return_tensors="np"
+            np_speech_inputs, padding=True, return_tensors="np", sampling_rate=dict_no_dither["sampling_rate"]
         ).input_features
         input_features_dither = feature_extractor_dither(
-            np_speech_inputs, padding=True, return_tensors="np"
+            np_speech_inputs, padding=True, return_tensors="np", sampling_rate=dict_dither["sampling_rate"]
         ).input_features
 
         # test there is a difference between features (there's added noise to input signal)
@@ -176,7 +176,7 @@ class Speech2TextFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unitt
         self.assertTrue(np.abs(diff).mean() > 1e-5)
         # features are not too different
         self.assertTrue(np.abs(diff).mean() <= 1e-3)
-        self.assertTrue(np.abs(diff).max() <= 1e-2)
+        self.assertTrue(np.abs(diff).max() <= 5e-2)
 
     def test_cepstral_mean_and_variance_normalization(self):
         feature_extractor = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict())

tests/models/whisper/test_feature_extraction_whisper.py

@@ -219,10 +219,10 @@ class WhisperFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest.
 
         # compute features
         input_features_no_dither = feature_extractor_no_dither(
-            np_speech_inputs, padding=True, return_tensors="np"
+            np_speech_inputs, padding=True, return_tensors="np", sampling_rate=dict_no_dither["sampling_rate"]
         ).input_features
         input_features_dither = feature_extractor_dither(
-            np_speech_inputs, padding=True, return_tensors="np"
+            np_speech_inputs, padding=True, return_tensors="np", sampling_rate=dict_dither["sampling_rate"]
         ).input_features
 
         # test there is a difference between features (there's added noise to input signal)
@@ -232,7 +232,7 @@ class WhisperFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest.
         self.assertTrue(np.abs(diff).mean() > 1e-6)
         # features are not too different
         self.assertTrue(np.abs(diff).mean() <= 1e-4)
-        self.assertTrue(np.abs(diff).max() <= 1e-3)
+        self.assertTrue(np.abs(diff).max() <= 5e-3)
 
     @require_torch
     def test_double_precision_pad(self):