Fix batch consistency test for InverseBarkScale (#3322)

Summary:
The batch consistency test function should call `InverseBarkScale` instead of `InverseMelScale`.

Pull Request resolved: https://github.com/pytorch/audio/pull/3322

Reviewed By: mthrok

Differential Revision: D45691769

Pulled By: nateanl

fbshipit-source-id: 4a1ed80c4a56c3a847a49a8d02f8b5cbe4f09045

test/torchaudio_unittest/prototype/transforms/batch_consistency_test.py

@@ -35,7 +35,7 @@ class BatchConsistencyTest(TorchaudioTestCase):
         n_barks = 32
         n_stft = 5
         bark_spec = torch.randn(3, 2, n_barks, 32) ** 2
-        transform = transforms.InverseMelScale(n_stft, n_barks)
+        transform = T.InverseBarkScale(n_stft, n_barks)
 
         # Because InverseBarkScale runs SGD on randomly initialized values so they do not yield
         # exactly same result. For this reason, tolerance is very relaxed here.