音频文件转换频谱图

音频文件转换频谱图，并保存为.jpg图片和.npy文件

import glob
import matplotlib.pyplot as plt
import numpy as np
from numpy import save
import os
import librosa

def save_img_and_npy(ad_path, save_path=None):

    sep = os.sep
    path_template = ad_path + sep + "*.wav"

    if save_path is None:
        save_path = "result" + sep + os.path.split(ad_path)[-1]

    npy_path=save_path+sep+"npy"

    img_path = save_path + sep + "image"

    if not os.path.exists(npy_path):
        os.makedirs(npy_path)

    if not os.path.exists(img_path):
        os.makedirs(img_path)

    hl = 512
    hi = 100
    wi = 384
    fmax = 1200

    for ad in glob.glob(path_template):
        name = os.path.split(ad)[-1].split(".")[0]
        npy_file = npy_path + os.sep + name+".npy"
        img_file = img_path + os.sep + name+".jpg"
        # 加载音频,return:数据、采样率
        y, sr = librosa.load(ad, sr=None)
        # # 展示波形图
        # plt.figure(figsize=(14, 5))
        # librosa.display.waveshow(y, sr=sr)

        S = librosa.feature.melspectrogram(y=y, sr=sr, n_mels=hi, fmax=fmax, n_fft=2048, hop_length=hl)
        # S = np.log(S + 1e-10)
        S_dB = librosa.power_to_db(S, ref=np.max)
        # print(S_dB.shape)
        save(npy_file,S_dB)
        librosa.display.specshow(S_dB, x_axis='time', y_axis='mel', sr=sr, fmax=fmax)

        plt.savefig(img_file)
        ##显示
        # plt.show()
        # plt.close()


if __name__ == '__main__':
    #.wav文件的文件夹路径
    p_list = [""]
    for p in p_list:
        save_img_and_npy(p)