Update app.py

app.py

@@ -1,12 +1,17 @@
-# Streamlit app code
 import streamlit as st
-from transformers import pipeline
+from transformers import AutoModelForAudioClassification, AutoFeatureExtractor, pipeline
 import os
-import tensorflow as tf
 
 # Title of the Streamlit app
 st.title('Music Genre Classification')
 
+# Load the model and feature extractor
+model = AutoModelForAudioClassification.from_pretrained("sandychoii/distilhubert-finetuned-gtzan-audio-classification")
+feature_extractor = AutoFeatureExtractor.from_pretrained("sandychoii/distilhubert-finetuned-gtzan-audio-classification")
+
+# Create the audio classification pipeline
+pipe = pipeline(task="audio-classification", model=model, feature_extractor=feature_extractor)
+
 # File uploader to upload an audio file
 audio_file = st.file_uploader("Upload an audio file", type=["wav", "mp3", "flac"])
 
@@ -15,25 +20,23 @@ if audio_file is not None:
     if not os.path.exists("temp_audio"):
         os.makedirs("temp_audio")
     
-    with open(os.path.join("temp_audio", audio_file.name), "wb") as f:
+    temp_path = os.path.join("temp_audio", audio_file.name)
+    with open(temp_path, "wb") as f:
         f.write(audio_file.getbuffer())
     
-    # Use the uploaded audio file path
-    audio_path = os.path.join("temp_audio", audio_file.name)
-
     # Display a loading message
     st.text("Classifying the audio file... Please wait.")
-
-    # Initialize the audio classification pipeline
-    pipe = pipeline("audio-classification", model="sandychoii/distilhubert-finetuned-gtzan-audio-classification")
     
     # Perform classification
-    result = pipe(audio_path)
-
+    result = pipe(temp_path)
+    
     # Display the results
     st.subheader("Classification Results:")
     for label in result:
         st.write(f"**Genre**: {label['label']} | **Confidence**: {label['score']:.4f}")
-
+    
     # Option to play the uploaded audio
     st.audio(audio_file)
+
+    # Clean up the temporary file
+    os.remove(temp_path)