[Fbsync] Update pitch shift batch consistency test (pytorch#1700)

* Reduce length of waveform in pitch_shift batch_consistency test

Summary: To address the test failure in T96406395

Reviewed By: carolineechen

Differential Revision: D30163741

fbshipit-source-id: f88d86b3da7b1ee52518934567b0b0a62700ee58

* Fix batch consistency test in transforms

Summary: The stress test still fails. Add n_fft to address it.

Reviewed By: mthrok

Differential Revision: D30218279

fbshipit-source-id: 7858efd3e5ac0073193a7883fd314486efc73814

Co-authored-by: Zhaoheng Ni <[email protected]>

Author: hwangjeff

test/torchaudio_unittest/transforms/batch_consistency_test.py

@@ -211,12 +211,12 @@ def test_batch_spectral_centroid(self):
 
     def test_batch_pitch_shift(self):
         sample_rate = 8000
-        n_steps = 4
-        waveform = common_utils.get_whitenoise(sample_rate=sample_rate)
+        n_steps = -2
+        waveform = common_utils.get_whitenoise(sample_rate=sample_rate, duration=0.05)
 
         # Single then transform then batch
-        expected = torchaudio.transforms.PitchShift(sample_rate, n_steps)(waveform).repeat(3, 1, 1)
+        expected = torchaudio.transforms.PitchShift(sample_rate, n_steps, n_fft=400)(waveform).repeat(3, 1, 1)
 
         # Batch then transform
-        computed = torchaudio.transforms.PitchShift(sample_rate, n_steps)(waveform.repeat(3, 1, 1))
+        computed = torchaudio.transforms.PitchShift(sample_rate, n_steps, n_fft=400)(waveform.repeat(3, 1, 1))
         self.assertEqual(computed, expected)