neural-amp-modeler

commit adee42ee3f7f9a4aad345c0830998fd8782a5d25
parent 213b72b70917e8acb655bed899b34d509a6448c0
Author: Steven Atkinson <[email protected]>
Date:   Mon, 29 May 2023 14:35:22 -0700

Fix AudioShapeMismatchError mixed up arguments (#258)


Diffstat:
M
nam/data.py
 | 
6
+++---
M
tests/test_nam/test_data.py
 | 
27
+++++++++++++++++++++++++++

2 files changed, 30 insertions(+), 3 deletions(-)
diff --git a/nam/data.py b/nam/data.py
@@ -91,10 +91,10 @@ def wav_to_np(
     if required_shape is not None:
         if arr_premono.shape != required_shape:
             raise AudioShapeMismatchError(
-                arr_premono.shape,
-                required_shape,
+                required_shape,  # Expected
+                arr_premono.shape,  # Actual
                 f"Mismatched shapes. Expected {required_shape}, but this is "
-                f"{arr_premono.shape}!",
+                f"{arr_premono.shape}!"
             )
         # sampwidth fine--we're just casting to 32-bit float anyways
     arr = arr_premono[:, 0]
diff --git a/tests/test_nam/test_data.py b/tests/test_nam/test_data.py
@@ -4,6 +4,8 @@
 
 import math
 from enum import Enum
+from pathlib import Path
+from tempfile import TemporaryDirectory
 from typing import Tuple
 
 import numpy as np
@@ -241,5 +243,30 @@ class TestDataset(object):
         return x_out, y_out
 
 
+def test_audio_mismatch_shapes_in_order():
+    """
+    https://github.com/sdatkinson/neural-amp-modeler/issues/257
+    """
+    x_samples, y_samples = 5, 7
+    num_channels = 1
+
+    x, y = [np.zeros((n, num_channels)) for n in (x_samples, y_samples)]
+    
+    with TemporaryDirectory() as tmpdir:
+        y_path = Path(tmpdir, "y.wav")
+        data.np_to_wav(y, y_path)
+        f = lambda: data.wav_to_np(y_path, required_shape=x.shape)
+    
+        with pytest.raises(data.AudioShapeMismatchError) as e:
+            f()
+
+        try:
+            f()
+            assert False, "Shouldn't have succeeded!"
+        except data.AudioShapeMismatchError as e:
+            # x is loaded first; we expect that y matches.
+            assert e.shape_expected == (x_samples, num_channels)
+            assert e.shape_actual == (y_samples, num_channels)
+
 if __name__ == "__main__":
     pytest.main()