diff --git a/src/infer.py b/src/infer.py
index 1726fd8..01b0969 100644
--- a/src/infer.py
+++ b/src/infer.py
@@ -5,7 +5,7 @@
 from torch.nn.utils.rnn import pad_sequence
 from torchdiffeq import odeint
 from safetensors.torch import load_file
-import IPython.display as ipd
+from pathlib import Path
 
 # Import F5-TTS modules
 from f5_tts.model import CFM, UNetT, DiT
@@ -126,7 +126,16 @@ def _setup_mel_spec(self):
         
     def _setup_vocoder(self):
         """Initialize vocoder."""
-        self.vocos = load_vocoder(is_local=False, local_path="")
+        # Provide path via ENV to a local copy of the model (otherwise it will be downloaded from HF):
+        # https://huggingface.co/charactr/vocos-mel-24khz
+        model_dir = Path(os.environ.get("MODEL_DIR_VOCODER", ""))
+        required_files = ["config.yaml", "pytorch_model.bin"]
+
+        if model_dir.is_dir() and all((model_dir / f).is_file() for f in required_files):
+            self.vocos = load_vocoder(is_local=True, local_path=model_dir)
+        else:
+            self.vocos = load_vocoder(is_local=False, local_path="")
+
         self.vocos = self.vocos.to(self.device)
         
     def _setup_duration_predictor(self, checkpoint_path):