test for speech directivity

pyroomacoustics/tests/test_speech_source_directivities.py

@@ -0,0 +1,76 @@
+import pyroomacoustics as pra
+import numpy as np
+from pyroomacoustics.directivities import (
+    DirectivityPattern,
+    DirectionVector,
+    SpeechDirectivity,
+)
+from unittest import TestCase
+
+
+
+def predict_rir_speech_source(speech_direction):
+    """
+    Predict RIR when speech is facing in a particular direction.
+    
+    Parameters
+    ----------
+    speech_direction: float
+        the direction that the speech source is facing
+
+    Returns
+    -------
+    :np.ndarray ``(float, [float, float])``
+        This function returns ``(m, [le, ue])`` and the confidence interval is ``[m-le, m+ue]``.
+    """
+
+    scat_test = {
+        "coeffs": [0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5],
+        "center_freqs": [125, 250, 500, 1000, 2000, 4000, 8000],
+    }
+    abs_test = {
+        "coeffs": [0.8, 0.8, 0.8, 0.8, 0.8, 0.8, 0.8],
+        "center_freqs": [125, 250, 500, 1000, 2000, 4000, 8000],
+    }
+    material = pra.Material(abs_test, scat_test)
+
+    # create room
+    room = pra.ShoeBox(
+        p=[7, 5, 5],
+        materials=material,
+        fs=16000,
+        max_order=1,
+    )
+
+    # define source with speech directivity
+    PATTERN = DirectivityPattern.FIGURE_EIGHT
+    ORIENTATION = DirectionVector(azimuth=speech_direction, colatitude=90, degrees=True)
+
+    directivity = SpeechDirectivity(orientation=ORIENTATION)
+
+    # add source with speech directivity
+    room.add_source([2.5, 2.5, 2.5], directivity=directivity)
+
+    # add microphone on axis with speech source
+    room.add_microphone([5, 2.5, 2.5])
+
+    # compute rir
+    room.compute_rir()
+
+    return room.rir[0][0]
+
+class TestSourceDirectivity(TestCase):
+    def test_speech_direction(self):
+        energy_0_deg = np.mean(predict_rir_speech_source(speech_direction=0)**2)
+        energy_180_deg = np.mean(predict_rir_speech_source(speech_direction=180)**2)
+
+        self.assertTrue(energy_0_deg > energy_180_deg)
+
+if __name__ == "__main__":
+    energy_0_deg = np.mean(predict_rir_speech_source(azimuth=0)**2)
+    energy_180_deg = np.mean(predict_rir_speech_source(azimuth=180)**2)
+    print()
+    print("-" * 40)
+    print(f"mean energy when speech facing towards mic {10 *np.log10(energy_0_deg):.0f} dB")
+    print(f"mean energy when speech facing away from mic {10 *np.log10(energy_180_deg):.0f} dB")
+    print("-" * 40)