CNNで好きなアーティストを分類させてAIに良さを伝えたい

import librosa
from pydub import AudioSegment

artist = ["AAA", "UNISON_SQUARE_GARDEN", "嵐", "NEWS", "SEKAI_NO_OWARI"]

i in range(len(artist)):
    file_index = 0
    pre_time = 0
    rea_time = 10000

    for j in range(10): 
        file_dir = "./music_data/" + artist[i] + "/raw_data/raw_" + str(j) + ".wav"
        music, fs = librosa.audio.load(file_dir)
        play_time = int(librosa.samples_to_time(len(music), fs)/10)
        
        for k in range(play_time):
            audio_data = AudioSegment.from_file(file_dir, format="wav")
            split_audio = audio_data[pre_time:rea_time]
            pre_time += 10000
            rea_time += 10000
            
            split_audio.export("./music_data/" + artist[i] + "/split_data/" + str(file_index) + ".wav", format = "wav")
            
            file_index += 1

for i in range(len(artist)):
    for j in range(split_num[i]):
        split_file_dir = "./music_data/" + artist[i] + "/split_data/" + str(j) + ".wav"
        export_dir = "./music_data/" + artist[i] + "/spectrum/" + str(j) + ".jpg"
        audio_data, sr = librosa.load(split_file_dir)
        S = librosa.feature.melspectrogram(audio_data, sr = sr, n_mels = 128)
        log_S = librosa.power_to_db(S)
        
        plt.figure(figsize=(8, 8))
        librosa.display.specshow(y=audio_data, sr=sr, hop_length = 2068)
        plt.savefig(export_dir)

from keras.utils.np_utils import to_categorical
improt cv2
import numpy as np

spe_dir = "./music_data/AAA/spectrum/"
X_train=[]
y_train=[]

for i in range(266):
    img = cv2.imread(spe_dir + str(i) + ".png")
    b,g,r = cv2.split(img)
    img = cv2.merge([r,g,b])
    X_train.append(img)
    y_train.append(0)

X_train = np.array(X_train)
y_train = np.array(y_train)
y_train = y_train.T
y_train = to_categorical(y_train)

from keras.layers import Activation, Conv2D, Dense, Flatten, MaxPooling2D
from keras.models import Sequential, load_model
from keras.utils.np_utils import to_categorical
from keras import optimizers
from keras.layers import Dense, Dropout, Flatten, Input, BatchNormalization
from keras.models import Model, Sequential

model = Sequential()
model.add(Conv2D(input_shape=(576, 576, 3), filters=32, kernel_size=(2, 2), strides=(1, 1), padding="same"))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Conv2D(filters=32, kernel_size=(2,2), strides=(1,1), padding="same"))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Conv2D(filters=32, kernel_size=(2,2), strides=(1,1), padding="same"))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Flatten())
model.add(Dense(256))
model.add(Activation("relu"))
model.add(Dense(128))
model.add(Activation('relu'))
model.add(BatchNormalization())
model.add(Dropout(0.5))
model.add(Dense(5))
model.add(Activation('softmax'))

model.compile(optimizer='Adam', loss='categorical_crossentropy',metrics=['accuracy'])
model.fit(X_train, y_train, batch_size=32, epochs=10)

score = model.evaluate(X_test, y_test, verbose=1)
print()
print("Test loss:", score[0])
print("Test accuracy:", score[1])