init

app.py

@@ -69,7 +69,7 @@ def transcribe(inputs: str, prompt):
         inputs = f.read()
     inputs = ffmpeg_read(inputs, pipe.feature_extractor.sampling_rate)
     array_pad = np.zeros(int(pipe.feature_extractor.sampling_rate * 0.5))
-    inputs = np.concatenate([inputs, inputs, inputs])
+    inputs = np.concatenate([array_pad, inputs, array_pad])
     inputs = {"array": inputs, "sampling_rate": pipe.feature_extractor.sampling_rate}
     return get_prediction(inputs, prompt)