set default align target param to MFCC

Author: stefantaubert

README.md

@@ -33,11 +33,11 @@ from mel_cepstral_distance import compare_audio_files
 
 mcd, penalty = compare_audio_files(
   'examples/GT.wav',
-  'examples/WaveGlow.wav',
+  'examples/Tacotron-2.wav',
 )
 
 print(f'MCD: {mcd:.2f}, Penalty: {penalty:.4f}')
-# MCD: 4.03, Penalty: 0.0197
+# MCD: 7.45, Penalty: 0.1087
 ```
 
 ## Calculation

src/mel_cepstral_distance/api.py

@@ -398,7 +398,7 @@ def compare_audio_files(
   s: int = 1,
   D: int = 16,
   aligning: Literal["pad", "dtw"] = "dtw",
-  align_target: Literal["spec", "mel", "mfcc"] = "mel",
+  align_target: Literal["spec", "mel", "mfcc"] = "mfcc",
   remove_silence: Literal["no", "sig", "spec", "mel", "mfcc"] = "no",
   silence_threshold_A: Optional[float] = None,
   silence_threshold_B: Optional[float] = None,
@@ -649,7 +649,7 @@ def compare_amplitude_spectrograms(
   s: int = 1,
   D: int = 16,
   aligning: Literal["pad", "dtw"] = "dtw",
-  align_target: Literal["spec", "mel", "mfcc"] = "spec",
+  align_target: Literal["spec", "mel", "mfcc"] = "mfcc",
   remove_silence: Literal["no", "spec", "mel", "mfcc"] = "no",
   silence_threshold_A: Optional[float] = None,
   silence_threshold_B: Optional[float] = None,
@@ -879,7 +879,7 @@ def compare_mel_spectrograms(
   s: int = 1,
   D: int = 16,
   aligning: Literal["pad", "dtw"] = "dtw",
-  align_target: Literal["mel", "mfcc"] = "mel",
+  align_target: Literal["mel", "mfcc"] = "mfcc",
   remove_silence: Literal["no", "mel", "mfcc"] = "no",
   silence_threshold_A: Optional[float] = None,
   silence_threshold_B: Optional[float] = None,

src/mel_cepstral_distance_tests/api_tests/test_compare_amplitude_spectrograms.py

@@ -376,14 +376,17 @@ def test_no_freq_bins_raises_error() -> None:
 
 
 def test_invalid_radius_raises_error() -> None:
-  with pytest.raises(ValueError):
+  with pytest.raises(
+    ValueError, match="dtw_radius must be None or greater than or equal to 1"
+  ):
     compare_amplitude_spectrograms(
       get_X_km_A(),
       get_X_km_B(),
       22050,
       samples_to_ms(512, 22050),
       dtw_radius=0,
       aligning="dtw",
+      align_target="spec",
     )
 
 

src/mel_cepstral_distance_tests/api_tests/test_compare_audio_files.py

@@ -559,7 +559,9 @@ def test_invalid_remove_silence_raises_error() -> None:
 
 
 def test_invalid_radius_raises_error() -> None:
-  with pytest.raises(ValueError):
+  with pytest.raises(
+    ValueError, match="dtw_radius must be None or greater than or equal to 1"
+  ):
     compare_audio_files(AUDIO_A, AUDIO_B, aligning="dtw", dtw_radius=0)
 
 

src/mel_cepstral_distance_tests/api_tests/test_compare_mel_spectrograms.py

@@ -131,8 +131,12 @@ def test_same_spec_returns_zero() -> None:
 
 
 def test_invalid_radius_raises_error() -> None:
-  with pytest.raises(ValueError):
-    compare_mel_spectrograms(get_X_kn_A(), get_X_kn_B(), aligning="dtw", dtw_radius=0)
+  with pytest.raises(
+    ValueError, match="dtw_radius must be None or greater than or equal to 1"
+  ):
+    compare_mel_spectrograms(
+      get_X_kn_A(), get_X_kn_B(), aligning="dtw", align_target="mel", dtw_radius=0
+    )
 
 
 def test_removing_silence_from_mel_too_hard_returns_nan_nan() -> None: