简单的音频能量和rmse计算方法

Zero_to_zero1234

已于 2022-08-15 01:02:09 修改

阅读量1.6k

点赞数

分类专栏：语音合成深度学习文章标签：音视频 python 开发语言

于 2022-07-15 16:53:17 首次发布

本文链接：https://blog.csdn.net/suiyueruge1314/article/details/125808228

版权

深度学习同时被 2 个专栏收录

112 篇文章 5 订阅

订阅专栏

语音合成

5 篇文章 4 订阅

订阅专栏

import wave
import numpy as np
import matplotlib.pyplot as pl
import librosa
import numpy as np
 
framerate = 22050
wav_path = "xxx/LJ001-0001.wav"


# 计算每一帧的能量 256个采样点为一帧
def calEnergy(x) :
    hop_length = 256
    frame_length = 512
    energy = np.array([
        sum(abs(x[i:i+frame_length]**2))
        for i in range(0, len(x), hop_length)
    ])

    rmse = librosa.feature.rms(x, frame_length=frame_length, hop_length=hop_length, center=True)

    return energy, rmse[0]

x, sr = librosa.load(wav_path)

energy, rmse = calEnergy(x)
 
 
time = np.arange(0, len(x)) * (1.0 / framerate)
time2 = np.arange(0, len(energy)) * (len(x)/len(energy) / framerate)
pl.subplot(311)
pl.plot(time, x)
pl.ylabel("Amplitude")
pl.subplot(312)
pl.plot(time2, energy)
pl.ylabel("short energy")
pl.xlabel("time (seconds)")

pl.subplot(313)
pl.plot(time2, rmse)
pl.ylabel("short rmse")
pl.xlabel("time (seconds)")

pl.show()

pl.savefig("./engergy.png")