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

audio_utils improvements (#21998)

Browse Source 
* silly change to allow making a PR

* clean up doc comments

* simplify hertz_to_mel and mel_to_hertz

* fixup

* clean up power_to_db

* also add amplitude_to_db

* move functions

* clean up mel_filter_bank

* fixup

* credit librosa & torchaudio authors

* add unit tests

* tests for power_to_db and amplitude_to_db

* add mel_filter_bank tests

* rewrite STFT

* add convenience spectrogram function

* missing transpose

* fewer transposes

* add integration test to M-CTC-T

* frame length can be either window or FFT length

* rewrite stft API

* add preemphasis coefficient

* move argument

* add log option to spectrogram

* replace M-CTC-T feature extractor

* fix api thing

* replace whisper STFT

* replace whisper mel filters

* replace tvlt's stft

* allow alternate window names

* replace speecht5 stft

* fixup

* fix integration tests

* fix doc comments

* remove manual FFT length calculation

* fix docs

* go away, deprecation warnings

* combine everything into spectrogram function

* add deprecated functions back

* fixup
This commit is contained in:
Matthijs Hollemans
2023-05-09 15:10:17 +02:00
committed by GitHub
parent 431b04d8c4
commit 7f91950901
14 changed files with 1356 additions and 615 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/models/speecht5/test_feature_extraction_speecht5.py
View File

@@ -395,7 +395,8 @@ class SpeechT5FeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest
input_speech = self._load_datasamples(1)
feature_extractor = SpeechT5FeatureExtractor()
input_values = feature_extractor(input_speech, return_tensors="pt").input_values
self.assertTrue(torch.allclose(input_values[0, :30], EXPECTED_INPUT_VALUES, atol=1e-4))
self.assertEquals(input_values.shape, (1, 93680))
self.assertTrue(torch.allclose(input_values[0, :30], EXPECTED_INPUT_VALUES, atol=1e-6))
def test_integration_target(self):
# fmt: off
@@ -410,4 +411,5 @@ class SpeechT5FeatureExtractionTest(SequenceFeatureExtractionTestMixin, unittest
input_speech = self._load_datasamples(1)
feature_extractor = SpeechT5FeatureExtractor()
input_values = feature_extractor(audio_target=input_speech, return_tensors="pt").input_values
self.assertEquals(input_values.shape, (1, 366, 80))
self.assertTrue(torch.allclose(input_values[0, 0, :30], EXPECTED_INPUT_VALUES, atol=1e-4))