Update audio_processing/A2T.py

audio_processing/A2T.py

@@ -14,7 +14,7 @@ class A2T:
         self.mic = mic
 
     def __generate_text(self, inputs, task: Optional[str] = None) -> str:
-        if inputs is  None:
+        if inputs is None:
             raise Exception("Inputs is None")
 
         transcribed_text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)["text"]
@@ -22,11 +22,19 @@ class A2T:
 
     @staticmethod
     def __preprocess(raw: bytes) -> np.ndarray:
-        print(f"Raw type : {type(raw)}")
-        chunk = io.BytesIO(raw)
-        audio, sample_rate = librosa.load(chunk, sr=16000)
-        print(f"Sample rate : {sample_rate}")
-        return audio
+        print(f"Raw type: {type(raw)}")
+
+        if not isinstance(raw, bytes):
+            raise ValueError("Expected raw audio data as bytes")
+            
+        try:
+            chunk = io.BytesIO(raw)
+            print(f"Chunk type: {type(chunk)}")
+            audio, sample_rate = librosa.load(chunk, sr=16000)
+            print(f"Sample rate : {sample_rate}")
+            return audio
+        except Exception as e:
+            print(f"Error loading audio: {e}")
  
     def predict(self) -> str:
         try:
@@ -40,7 +48,7 @@ class A2T:
             if isinstance(audio, np.ndarray):
                 return self.__generate_text(inputs=audio, task=TASK)
             else:
-                raise Exception("Audio is not np array")
+                raise ValueError("Audio is not np array")
                 
         except Exception as e:
-            return f"Oops some kinda error : {e}"
+            print(f"Oops some kinda error : {e}")