Update app.py

app.py

@@ -64,7 +64,6 @@ model = WhisperForConditionalGeneration.from_pretrained(model_id)
 model.eval()
 model.to(device)
 
-sampling_rate = processor.feature_extractor.sampling_rate
 
 bos_token_id = processor.tokenizer.all_special_ids[-106]
 decoder_input_ids = torch.tensor([bos_token_id]).to(device)
@@ -84,7 +83,7 @@ def detect_language(sentence):
     return language, probability.item()
 
 
-def process_audio_file(file):
+def process_audio_file(file, sampling_rate):
     with open(file, "rb") as f:
         inputs = f.read()
 
@@ -107,7 +106,10 @@ def transcribe(Microphone, File_Upload):
     else:
         file = File_Upload
 
-    audio_data = process_audio_file(file)
+    sampling_rate = processor.feature_extractor.sampling_rate
+
+
+    audio_data = process_audio_file(file, sampling_rate)
 
     input_features = processor(audio_data, return_tensors="pt").input_features