SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

is_batched fix for remaining 2-D numpy arrays (#23309)

Browse Source 
* Fix is_batched code to allow 2-D numpy arrays for audio

* Tests

* Fix typo

* Incorporate comments from PR #23223
This commit is contained in:
LWprogramming
2023-05-23 11:37:35 -07:00
committed by GitHub
parent 6b7d6f848b
commit 3d57404464
14 changed files with 106 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
tests/models/speech_to_text/test_feature_extraction_speech_to_text.py
View File

@@ -136,6 +136,14 @@ class Speech2TextFeatureExtractionTest(SequenceFeatureExtractionTestMixin, unitt
for enc_seq_1, enc_seq_2 in zip(encoded_sequences_1, encoded_sequences_2):
self.assertTrue(np.allclose(enc_seq_1, enc_seq_2, atol=1e-3))
# Test 2-D numpy arrays are batched.
speech_inputs = [floats_list((1, x))[0] for x in (800, 800, 800)]
np_speech_inputs = np.asarray(speech_inputs)
encoded_sequences_1 = feature_extractor(speech_inputs, return_tensors="np").input_features
encoded_sequences_2 = feature_extractor(np_speech_inputs, return_tensors="np").input_features
for enc_seq_1, enc_seq_2 in zip(encoded_sequences_1, encoded_sequences_2):
self.assertTrue(np.allclose(enc_seq_1, enc_seq_2, atol=1e-3))
def test_cepstral_mean_and_variance_normalization(self):
feature_extractor = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict())
speech_inputs = [floats_list((1, x))[0] for x in range(800, 1400, 200)]