diff --git a/ivas_processing_scripts/generation/generate_masa_items.py b/ivas_processing_scripts/generation/generate_masa_items.py
index 65b29761261af5a609e67713878e0415a4739816..9d71d7c4d2ce6662bb7634fac101cd998e41e23f 100644
--- a/ivas_processing_scripts/generation/generate_masa_items.py
+++ b/ivas_processing_scripts/generation/generate_masa_items.py
@@ -208,6 +208,7 @@ def generate_MASA_scene(
 
     # repeat for all source files
     offset = 0
+    y_int = None
     for i in range(N_inputs):
         # parse parameters from the scene description
         source_file = (
@@ -269,9 +270,10 @@ def generate_MASA_scene(
             )
             sys.exit(-1)
 
-        # initialize intermediate SBA object
-        y_int = audio.SceneBasedAudio(IR_fmt)
-        y_int.fs = cfg.fs
+        # initialize intermediate SBA object (based on the format of the IR file)
+        if y_int is None:
+            y_int = audio.SceneBasedAudio(IR_fmt)
+            y_int.fs = cfg.fs
 
         # read source file
         x = audio.fromfile("MONO", input_filename)
@@ -325,13 +327,15 @@ def generate_MASA_scene(
             delta_offset = source_shift - offset
             if delta_offset > 0:
                 # insert zeros to the output SBA signal to shift it right
-                audioarray.trim(
+                y_int.audio = audioarray.trim(
                     y_int.audio, y_int.fs, limits=[-delta_offset, 0], samples=True
                 )
                 offset = source_shift
             else:
                 # insert zeros to the new SBA source signal to shift it right
-                audioarray.trim(x.audio, x.fs, limits=[delta_offset, 0], samples=True)
+                x.audio = audioarray.trim(
+                    x.audio, x.fs, limits=[delta_offset, 0], samples=True
+                )
 
             # adjust the length of the audio source signal
             delta_length = len(x.audio) - len(y_int.audio)
diff --git a/ivas_processing_scripts/generation/generate_mc_items.py b/ivas_processing_scripts/generation/generate_mc_items.py
index bac49e0c8de1720c46334cb9d576426da279e06b..2b2da206b3b832ca821f6a2bb3ed0bad6729ff23 100644
--- a/ivas_processing_scripts/generation/generate_mc_items.py
+++ b/ivas_processing_scripts/generation/generate_mc_items.py
@@ -208,6 +208,7 @@ def generate_MC_scene(
 
     # repeat for all source files
     offset = 0
+    y_int = None
     for i in range(N_inputs):
         # parse parameters from the scene description
         source_file = (
@@ -270,8 +271,9 @@ def generate_MC_scene(
             sys.exit(-1)
 
         # initialize intermediate SBA object (based on the format of the IR file)
-        y_int = audio.SceneBasedAudio(IR_fmt)
-        y_int.fs = cfg.fs
+        if y_int is None:
+            y_int = audio.SceneBasedAudio(IR_fmt)
+            y_int.fs = cfg.fs
 
         # read source file
         x = audio.fromfile("MONO", input_filename)
@@ -325,13 +327,15 @@ def generate_MC_scene(
             delta_offset = source_shift - offset
             if delta_offset > 0:
                 # insert zeros to the output SBA signal to shift it right
-                audioarray.trim(
+                y_int.audio = audioarray.trim(
                     y_int.audio, y_int.fs, limits=[-delta_offset, 0], samples=True
                 )
                 offset = source_shift
             else:
                 # insert zeros to the new SBA source signal to shift it right
-                audioarray.trim(x.audio, x.fs, limits=[delta_offset, 0], samples=True)
+                x.audio = audioarray.trim(
+                    x.audio, x.fs, limits=[delta_offset, 0], samples=True
+                )
 
             # adjust the length of the audio source signal
             delta_length = len(x.audio) - len(y_int.audio)