Update app.py

app.py

@@ -50,12 +50,18 @@ import torch
 import streamlit as st
 from audio_recorder_streamlit import audio_recorder
 import numpy as np
-from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq
+#from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq
+# Load model directly
+from transformers import AutoProcessor, AutoModelForPreTraining
 
-def transcribe_audio(audio_bytes):
-    processor = AutoProcessor.from_pretrained("facebook/s2t-small-librispeech-asr")
-    model = AutoModelForSpeechSeq2Seq.from_pretrained("facebook/s2t-small-librispeech-asr")
+processor = AutoProcessor.from_pretrained("facebook/wav2vec2-base")
+model = AutoModelForPreTraining.from_pretrained("facebook/wav2vec2-base")
 
+def transcribe_audio(audio_bytes):
+ #   processor = AutoProcessor.from_pretrained("facebook/s2t-small-librispeech-asr")
+  #  model = AutoModelForSpeechSeq2Seq.from_pretrained("facebook/s2t-small-librispeech-asr")
+    processor = AutoProcessor.from_pretrained("facebook/wav2vec2-base")
+    model = AutoModelForPreTraining.from_pretrained("facebook/wav2vec2-base")
     # Convert audio bytes to numpy array
     audio_array = np.frombuffer(audio_bytes, dtype=np.int16)