Changes required to enable FOA on TAU rooms (#55)

* changes required to enable FOA

* version bump

* SELDnet23 can be trained and replicates cross-validation results on STARSS23 Fold4 using the data generated with this commit.

Author: iranroman

example_generation.py

@@ -9,7 +9,7 @@
     "datasets/rir_datasets"  # Directory containing Room Impulse Response (RIR) files
 )
 ROOM = "bomb_shelter"  # Initial room setting, change according to available rooms listed below
-FORMAT = "mic"  # Output format specifier
+FORMAT = "foa"  # Output format specifier: could be 'mic' or 'foa'
 N_EVENTS_MEAN = 15  # Mean number of foreground events in a soundscape
 N_EVENTS_STD = 6  # Standard deviation of the number of foreground events
 DURATION = 60.0  # Duration in seconds of each soundscape, customizable by the user

spatialscaper/core.py

@@ -69,15 +69,15 @@
 __ROOM_RIR_FILE__ = {
     "metu": "metu_sparg_em32.sofa",
     "arni": "arni_mic.sofa",
-    "bomb_shelter": "bomb_shelter_mic.sofa",
-    "gym": "gym_mic.sofa",
-    "pb132": "pb132_mic.sofa",
-    "pc226": "pc226_mic.sofa",
-    "sa203": "sa203_mic.sofa",
-    "sc203": "sc203_mic.sofa",
-    "se203": "se203_mic.sofa",
-    "tb103": "tb103_mic.sofa",
-    "tc352": "tc352_mic.sofa",
+    "bomb_shelter": "bomb_shelter_{fmt}.sofa",
+    "gym": "gym_{fmt}.sofa",
+    "pb132": "pb132_{fmt}.sofa",
+    "pc226": "pc226_{fmt}.sofa",
+    "sa203": "sa203_{fmt}.sofa",
+    "sc203": "sc203_{fmt}.sofa",
+    "se203": "se203_{fmt}.sofa",
+    "tb103": "tb103_{fmt}.sofa",
+    "tc352": "tc352_{fmt}.sofa",
 }
 
 
@@ -547,8 +547,10 @@ def get_room_irs_xyz(self):
         Returns:
             numpy.ndarray: An array of XYZ coordinates for the impulse response positions.
         """
+        if self.format == 'foa' and self.room in ['metu','arni']:
+            raise ValueError('"metu" and "arni" rooms are currently only supported in mic (tetrahedral) format. please check again soon.')
         room_sofa_path = os.path.join(
-            self.rir_dir, __SPATIAL_SCAPER_RIRS_DIR__, __ROOM_RIR_FILE__[self.room]
+            self.rir_dir, __SPATIAL_SCAPER_RIRS_DIR__, __ROOM_RIR_FILE__[self.room].format(fmt=self.format)
         )
         return load_pos(room_sofa_path, doas=False)
 
@@ -563,8 +565,10 @@ def get_room_irs_wav_xyz(self, wav=True, pos=True):
         Returns:
             tuple: A tuple containing the impulse responses, their sampling rate, and their XYZ positions.
         """
+        if self.format == 'foa' and self.room in ['metu','arni']:
+            raise ValueError('"metu" and "arni" rooms are currently only supported in mic (tetrahedral) format. please check again soon.')
         room_sofa_path = os.path.join(
-            self.rir_dir, __SPATIAL_SCAPER_RIRS_DIR__, __ROOM_RIR_FILE__[self.room]
+            self.rir_dir, __SPATIAL_SCAPER_RIRS_DIR__, __ROOM_RIR_FILE__[self.room].format(fmt=self.format)
         )
         all_irs, ir_sr, all_ir_xyzs = load_rir_pos(room_sofa_path, doas=False)
         ir_sr = ir_sr.data[0]

spatialscaper/version.py

@@ -3,4 +3,4 @@
 """Version info"""
 
 short_version = "0.1"
-version = "0.1.1"
+version = "0.1.2"