Update app.py

app.py

@@ -4,6 +4,9 @@ import torch
 import librosa
 import io
 import os
+from silero_vad import load_silero_vad, read_audio, get_speech_timestamps
+model = load_silero_vad()
+
 
 app = FastAPI()
 
@@ -31,6 +34,16 @@ async def transcribe_audio(file: UploadFile = File(...)):
         audio_bytes = await file.read()
         audio_array, sampling_rate = librosa.load(io.BytesIO(audio_bytes), sr=processor.feature_extractor.sampling_rate)
 
+        
+
+        speech_timestamps = get_speech_timestamps(
+          torch.from_numpy(audio_array),
+          model,
+          return_seconds=True,  # Return speech timestamps in seconds (default is samples)
+        )
+
+        print(speech_timestamps)
+
         inputs = processor(
             audio_array,
             return_tensors="pt",