Update asr.py

asr.py

@@ -1,6 +1,5 @@
 import librosa
-from transformers import Wav2Vec2ForCTC, AutoProcessor
-import torch
+from transformers import pipeline
 import logging
 
 # Set up logging
@@ -8,14 +7,11 @@ logging.basicConfig(level=logging.DEBUG)
 
 ASR_SAMPLING_RATE = 16_000
 
-MODEL_ID = "facebook/wav2vec2-large-960h-lv60-self"
-
 try:
-    processor = AutoProcessor.from_pretrained(MODEL_ID)
-    model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)
-    logging.info("ASR model and processor loaded successfully.")
+    pipe = pipeline("automatic-speech-recognition", model="facebook/mms-1b-all")
+    logging.info("ASR pipeline loaded successfully.")
 except Exception as e:
-    logging.error(f"Error loading ASR model or processor: {e}")
+    logging.error(f"Error loading ASR pipeline: {e}")
 
 def transcribe(audio):
     try:
@@ -25,17 +21,9 @@ def transcribe(audio):
         
         logging.info(f"Loading audio file: {audio}")
         audio_samples, _ = librosa.load(audio, sr=ASR_SAMPLING_RATE, mono=True)
-        inputs = processor(audio_samples, sampling_rate=ASR_SAMPLING_RATE, return_tensors="pt")
-
-        device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-        model.to(device)
-        inputs = inputs.to(device)
-
-        with torch.no_grad():
-            outputs = model(**inputs).logits
-
-        ids = torch.argmax(outputs, dim=-1)[0]
-        transcription = processor.decode(ids)
+        
+        # Process the audio with the pipeline
+        transcription = pipe(audio_samples, sampling_rate=ASR_SAMPLING_RATE)["text"]
         
         logging.info("Transcription completed successfully.")
         return transcription