Update asr.py

asr.py

@@ -68,8 +68,7 @@ model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)
 
 
 def transcribe(audio_data=None, lang="eng (English)"):
-
-    if not audio_data:
+    if audio_data is None or (isinstance(audio_data, np.ndarray) and audio_data.size == 0):
         return "<<ERROR: Empty Audio Input>>"
     
     if isinstance(audio_data, tuple):
@@ -80,11 +79,14 @@ def transcribe(audio_data=None, lang="eng (English)"):
             audio_samples = librosa.resample(
                 audio_samples, orig_sr=sr, target_sr=ASR_SAMPLING_RATE
             )
-    else:
+    elif isinstance(audio_data, np.ndarray):
+        # Assuming audio_data is already in the correct format
+        audio_samples = audio_data
+    elif isinstance(audio_data, str):
         # file upload
-        
-        if not isinstance(audio_data, str):
-            return "<<ERROR: Invalid Audio Input Instance: {}>>".format(type(audio_data))
+        audio_samples = librosa.load(audio_data, sr=ASR_SAMPLING_RATE, mono=True)[0]
+    else:
+        return f"<<ERROR: Invalid Audio Input Instance: {type(audio_data)}>>"
         audio_samples = librosa.load(audio_data, sr=ASR_SAMPLING_RATE, mono=True)[0]
 
     lang_code = lang.split()[0]