Add tests

src/sse/music_v2.py

@@ -55,7 +55,9 @@ def fit_transform(
             X_freq.conj().reshape(-1, self.N_ch, 1) @ X_freq.reshape(-1, 1, self.N_ch)
         ) / (X_freq.shape[0] / self.fs)
 
-        self.eigval, self.eigvec = np.linalg.eig(self.R)
+        # self.eigval, self.eigvec = np.linalg.eig(self.R)
+        self.eigvec, self.eigval, _ = np.linalg.svd(self.R)
+
         self.minvec = self.eigvec[:, :, -1].reshape(-1, self.N_ch, 1)
 
         self.thetas = np.linspace(-np.pi / 2, np.pi / 2, self.N_theta)

tests/test_music.py

@@ -5,12 +5,40 @@
 from sse.music_v2 import MusicSoundSourceLocator
 
 
+def make_dummy_signals(
+    theta=15.0,
+    fs=16000,
+    N_fft=128,
+    c=340,
+    d=1.0,
+) -> np.ndarray:
+    tdoa = d * np.sin(np.deg2rad(theta)) / c
+    # tdoa = d * np.cos(np.deg2rad(theta)) / c
+
+    width = N_fft // 2
+    # s = np.random.randn(N_fft + 2*width) # signal source.
+    ptdoa = np.round(tdoa * fs).astype("i")  # point of TDOA.
+    t = np.linspace(0, N_fft + 2 * width - 1, N_fft + 2 * width) / fs  # base time.
+    t1 = t[width + ptdoa : width + N_fft + ptdoa]
+    t2 = t[width : width + N_fft]
+    x1 = np.sin(2 * np.pi * 500 * t1)[:, None]
+    x2 = np.sin(2 * np.pi * 500 * t2)[:, None]
+    x = np.c_[x1, x2]
+    return x
+
+
 class TestMusicSoundSourceLocator:
     def test_fit_transform(self):
         self.locator = MusicSoundSourceLocator(
-            fs=16000,
-            d=1.0,
+            fs=48000,
+            d=0.1,
+            N_theta=180,
+        )
+        X = make_dummy_signals(
+            theta=40.0,
+            fs=48000,
+            N_fft=2048,
+            d=0.1,
         )
-        X = np.random.randn(1000, 2)
-        out = self.locator.fit_transform(X=X)
-        print(out)
+        predicted_theta = self.locator.fit_transform(X=X)
+        np.testing.assert_allclose(predicted_theta, 42.737430)