diff --git a/src/core/operators/audio_cnn_model/config.py b/src/core/operators/audio_cnn_model/config.py
index 00acb029..f7d93d6b 100644
--- a/src/core/operators/audio_cnn_model/config.py
+++ b/src/core/operators/audio_cnn_model/config.py
@@ -3,6 +3,7 @@
 # import numpy as np
 import csv
 from pathlib import Path
+import wget
 
 sample_rate = 32000
 
@@ -11,11 +12,13 @@
 # Download labels if not exist
 if not os.path.isfile(labels_csv_path):
     os.makedirs(os.path.dirname(labels_csv_path), exist_ok=True)
-    os.system(
-        'wget -O "{}" "http://storage.googleapis.com/us_audioset/youtube_corpus/v1/csv/class_labels_indices.csv"'.format(
-            labels_csv_path
-        )
-    )
+    # os.system(
+    #     'wget -O "{}" "http://storage.googleapis.com/us_audioset/youtube_corpus/v1/csv/class_labels_indices.csv"'.format(
+    #         labels_csv_path
+    #     )
+    # )
+    dl_path = "http://storage.googleapis.com/us_audioset/youtube_corpus/v1/csv/class_labels_indices.csv"
+    wget.download(dl_path, out=labels_csv_path)
 
 # Load label
 with open(labels_csv_path, "r") as f:
diff --git a/src/core/operators/audio_cnn_model/inference.py b/src/core/operators/audio_cnn_model/inference.py
index baaa2541..fbfac149 100644
--- a/src/core/operators/audio_cnn_model/inference.py
+++ b/src/core/operators/audio_cnn_model/inference.py
@@ -6,6 +6,7 @@
 # import matplotlib.pyplot as plt
 import torch
 from pathlib import Path
+import wget
 
 from .pytorch_utils import move_data_to_device
 from .models import Cnn14  # , Cnn14_DecisionLevelMax
@@ -39,7 +40,8 @@ def __init__(self, model=None, checkpoint_path=None, device="cuda"):
         ):
             create_folder(os.path.dirname(checkpoint_path))
             zenodo_path = "https://github.com/tattle-made/feluda/releases/download/third-party-models/Cnn14_mAP.0.431.pth"
-            os.system('wget -O "{}" "{}"'.format(checkpoint_path, zenodo_path))
+            # os.system('wget -O "{}" "{}"'.format(checkpoint_path, zenodo_path))
+            wget.download(zenodo_path, out=checkpoint_path)
 
         # script_dir = os.path.dirname(os.path.abspath(__file__))
         # checkpoint_path = os.path.join(script_dir, 'panns_data', 'Cnn14_mAP=0.431.pth')