Update app.py

app.py

@@ -32,6 +32,15 @@ def get_melspec_feature(X, target_sr, frame_size, frame_shift_len, n_mels):
         melspec_feature.append(audio_melspec)
     return np.array(melspec_feature, dtype=np.float32)
 
+def process_data_for_conv2D(X, resize_shape=None):
+    X_conv2D = []
+    for sample in X:
+        sample = np.reshape(sample, newshape=(sample.shape[0], sample.shape[1], 1))
+        if resize_shape:
+            sample = resize(sample, output_shape=resize_shape)
+        X_conv2D.append(sample)
+    return np.array(X_conv2D, dtype=np.float32)
+
 
 def predict_genre(audio):
     sr, audio = audio
@@ -39,7 +48,7 @@ def predict_genre(audio):
     # audio = np.ndarray(audio)
     audio = audio.astype(np.float32)
     melspec = get_melspec_feature(audio, target_sr, frame_size, frame_shift_len, 128) 
-    melspec = melspec.reshape(-1, 431, 128, 1)
+    melspec = process_data_for_conv2D(melspec)
     prediction = model.predict(melspec)[0]
     return {genre_classes[i]: float(prediction[i]) for i in range(5)}