使用tensorflow、keras在python中加载GTZAN数据集

import os
import librosa
import numpy as np
from sklearn.model_selection import train_test_split
from keras.utils import to_categorical

def load_dataset(dataset_path):
    X = []
    y = []
    for root, dirs, files in os.walk(dataset_path):
        for file in files:
            file_path = os.path.join(root, file)
            class_name = os.path.basename(root)
            signal, sr = librosa.load(file_path, sr=None)
            X.append(signal)
            y.append(class_name)
    return X, y

dataset_path = 'path_to_dataset'
X, y = load_dataset(dataset_path)

# 将音频信号转换为定长的特征向量
max_length = 30 * 22050  # 30秒音频对应的采样点数
X_processed = []
for signal in X:
    if len(signal) < max_length:
        signal = np.pad(signal, (0, max_length - len(signal)), mode='constant')
    else:
        signal = signal[:max_length]
    X_processed.append(signal)

# 将类别标签进行one-hot编码
labels = np.unique(y)
label_dict = {label: i for i, label in enumerate(labels)}
y_encoded = [label_dict[label] for label in y]
y_encoded = to_categorical(y_encoded)

X_train, X_test, y_train, y_test = train_test_split(X_processed, y_encoded, test_size=0.2, random_state=42)