Update app.py

app.py

@@ -63,8 +63,18 @@ feature_extractor = Wav2Vec2FeatureExtractor.from_pretrained("facebook/wav2vec2-
 model = Wav2Vec2Model.from_pretrained("facebook/wav2vec2-base-960h")
 
 # Function to convert audio file into embeddings
-def get_audio_embedding(audio_path):
-    audio, sr = librosa.load(audio_path, sr=16000)
+from io import BytesIO
+import librosa
+
+# Updated function for Streamlit-compatible audio processing
+def get_audio_embedding(uploaded_file):
+    # Convert uploaded file to in-memory buffer
+    audio_bytes = BytesIO(uploaded_file.read())
+    
+    # Load audio using librosa from the BytesIO buffer
+    audio, sr = librosa.load(audio_bytes, sr=16000)
+    
+    # Extract features using the Wav2Vec2 model
     inputs = feature_extractor(audio, sampling_rate=sr, return_tensors="pt", padding=True)
     with torch.no_grad():
         embeddings = model(**inputs).last_hidden_state.mean(dim=1)