diff --git a/transcribe.py b/transcribe.py
index b29581b..1160bea 100644
--- a/transcribe.py
+++ b/transcribe.py
@@ -7,7 +7,7 @@ import torch
 from silero_vad import load_silero_vad, get_speech_timestamps
 
 # Parameters
-MODEL_PATH = "mlx-community/whisper-tiny.en-mlx" # MLX optimized model
+MODEL_PATH = "mlx-community/whisper-small.en-mlx" # MLX optimized small model
 CHANNELS = 1
 SAMPLERATE = 16000
 BLOCK_SIZE = 512 # Silero VAD prefers 512, 1024, or 1536