Create app.py

app.py

@@ -0,0 +1,113 @@
+import gradio as gr
+import librosa
+import numpy as np
+import tensorflow as tf
+import pickle
+from pydub import AudioSegment
+import os
+
+
+# === Load Model dan Label Encoder ===
+model = tf.keras.models.load_model('/content/drive/MyDrive/AI CHORD RECOGNITION/Final/final_model.h5')
+with open('/content/drive/MyDrive/AI CHORD RECOGNITION/Final/label_chord.pkl', 'rb') as f:
+    label_encoder = pickle.load(f)
+
+# === Konversi MP3 ke WAV ===
+def convert_mp3_to_wav(mp3_path):
+    sound = AudioSegment.from_mp3(mp3_path)
+    wav_path = mp3_path.replace('.mp3', '.wav')
+    sound.export(wav_path, format="wav")
+    return wav_path
+
+# === Konversi Audio ke Mel Spectrogram ===
+def audio_to_mel_spectrogram(y, sr):
+    y = librosa.util.normalize(y)
+    mel_spectrogram = librosa.feature.melspectrogram(y=y, sr=sr, n_fft=2048, hop_length=512, n_mels=128)
+    mel_spectrogram_db = librosa.power_to_db(mel_spectrogram, ref=np.max)
+    mel_spectrogram_db = (mel_spectrogram_db + 80) / 80  # Normalisasi ke 0-1
+    mel_spectrogram_db = tf.image.resize(mel_spectrogram_db[..., np.newaxis], (128, 128)).numpy()
+    mel_spectrogram_db = np.repeat(mel_spectrogram_db, 3, axis=-1)
+    return np.expand_dims(mel_spectrogram_db, axis=0)
+
+# === Prediksi Chord ===
+def predict_chords(audio_path):
+    if audio_path.endswith('.mp3'):
+        audio_path = convert_mp3_to_wav(audio_path)
+    
+    y, sr = librosa.load(audio_path, sr=22050)
+    duration = librosa.get_duration(y=y, sr=sr)
+    chords = []
+    previous_chord = None
+
+    for i in range(0, int(duration)):
+        start_sample = i * sr
+        end_sample = (i + 1) * sr
+        y_segment = y[start_sample:end_sample]
+        if len(y_segment) == 0:
+            continue
+
+        mel_spectrogram = audio_to_mel_spectrogram(y_segment, sr)
+        prediction = model.predict(mel_spectrogram)
+        predicted_index = np.argmax(prediction)
+        predicted_chord = label_encoder.classes_[predicted_index]
+        predicted_chord = predicted_chord.replace('_', '') 
+
+        if predicted_chord != previous_chord:
+            chords.append(predicted_chord)
+            previous_chord = predicted_chord
+
+    return f"Predicted Chords: {' - '.join(chords)}"
+
+# === Gradio Interface ===
+css = """
+body {
+    font-family: sans-serif;
+}
+.gradio-container {
+    border-radius: 10px;
+    box-shadow: 0 2px 5px rgba(0, 0, 0, 0.1);
+    padding: 20px;
+}
+.gr-button {
+    background-color: #4CAF50;
+    color: white;
+    border: none;
+    padding: 6px 12px;
+    font-size: 14px;
+    cursor: pointer;
+    border-radius: 5px;
+}
+.gr-button:hover {
+    background-color: #4a894c;
+}
+.description {
+    margin-top: 40px;
+    margin-bottom:-10px;
+}
+"""
+
+sample_audio_path = "/content/drive/MyDrive/AI CHORD RECOGNITION/Final/example.mp3"
+
+title = f"AI Chord Recognition"
+description = """
+<div class='description'>
+Upload an audio file (<strong>.mp3</strong> or <strong>.wav</strong>) and let the AI predict the chord progression. 
+You can also try my sample audio.
+</div>
+"""
+
+with gr.Blocks(css=css) as interface:
+    gr.Markdown(f"<h1 style='text-align: center;'>{title}</h1>")  
+    gr.Markdown("<h3 style='text-align: center;'>Hello! I'm Tigor Neilson Sinaga, with NIM 22.11.4725.<br>Welcome to my AI Chord Recognition project!</h3>")
+    gr.Markdown(f"<h3 style='text-align: center;'>{description}</p>")
+    
+    audio_input = gr.Audio(type="filepath", label="Upload Audio (MP3/WAV)")
+    use_sample_button = gr.Button("Use Sample Audio", size="sm")
+    predict_button = gr.Button("Predict Chords") # Changed line
+    output_text = gr.Textbox(label="Predicted Chords", lines=5, placeholder="Chord predictions will appear here...")
+    
+    use_sample_button.click(fn=lambda: sample_audio_path, inputs=[], outputs=audio_input)
+    predict_button.click(fn=predict_chords, inputs=audio_input, outputs=output_text, queue=True)    
+    gr.Markdown("<p style='text-align: center; font-size: 12px; color: grey;'>This project is still under development and has not yet reached high accuracy.</p>")
+
+interface.launch(share=True)