diff --git a/pyAudioAnalysis/audioTrainTest.py b/pyAudioAnalysis/audioTrainTest.py
index 32bea38c6..7b7c5c224 100644
--- a/pyAudioAnalysis/audioTrainTest.py
+++ b/pyAudioAnalysis/audioTrainTest.py
@@ -17,6 +17,33 @@
 import plotly.graph_objs as go
 import sklearn.metrics
 
+# Fix for Arbitrary code execution due to loading of untrusted pickled data
+import io
+import builtins
+import pickle
+
+safe_builtins = {
+    'range',
+    'complex',
+    'set',
+    'frozenset',
+    'slice',
+}
+
+class RestrictedUnpickler(pickle.Unpickler):
+
+    def find_class(self, module, name):
+        # Only allow safe classes from builtins
+        if module == "builtins" and name in safe_builtins:
+            return getattr(builtins, name)
+        """Forbid everything else"""
+        raise pickle.UnpicklingError("global '%s.%s' is forbidden" %
+                                     (module, name))
+
+def restricted_loads(s):
+    # Helper function analogous to pickle.loads()
+    return RestrictedUnpickler(io.BytesIO(s)).load()
+
 
 shortTermWindow = 0.050
 shortTermStep = 0.050
@@ -502,6 +529,7 @@ def feature_extraction_train_regression(folder_name, mid_window, mid_step,
 
 def load_model_knn(knn_model_name, is_regression=False):
     with open(knn_model_name, "rb") as fo:
+        restricted_loads(fo.read())
         features = cPickle.load(fo)
         labels = cPickle.load(fo)
         mean = cPickle.load(fo)
@@ -540,6 +568,7 @@ def load_model(model_name, is_regression=False):
                              is regression or not
     """
     with open(model_name + "MEANS", "rb") as fo:
+        restricted_loads(fo.read())
         mean = cPickle.load(fo)
         std = cPickle.load(fo)
         if not is_regression:
@@ -554,6 +583,7 @@ def load_model(model_name, is_regression=False):
     std = np.array(std)
 
     with open(model_name, 'rb') as fid:
+        restricted_loads(fid.read())
         svm_model = cPickle.load(fid)
 
     if is_regression: