一个DSP滤波器设计工具，可以设计多种FIR和IIR滤波器

# BUG 1024 import os import sys import wave import time import ui_cn_1 import matplotlib import numpy as np import matplotlib.pyplot as plt from scipy import signal from PyQt5.QtCore import * from scipy.io import wavfile from PyQt5.QtWidgets import * from PyQt5.QtMultimedia import * from QCandyUi import CandyWindow from PyQt5.QtGui import QTextCursor from matplotlib.backends.backend_qt5agg import FigureCanvasQTAgg as FigureCanvas from PyQt5.QtWidgets import QApplication, QMainWindow, QFileDialog, QComboBox, QCheckBox matplotlib.rcParams['font.family'] = 'STSong' matplotlib.use('Qt5Agg') # 处理函数 class ProcessFunction(object): # 归一化幅度 def Audio_TimeDomain(self, feature): # 时域 f = wave.open(feature.path, "rb") # 打开音频文件 params = f.getparams() nchannels, sampwidth, framerate, nframes = params[:4] # nchannels通道数 # sampwidth量化位数 # framerate采样频率 # nframes采样点数 str_data = f.readframes(nframes) # 返回Bytes对象表示的采样点数设置的音频 f.close() # 关闭音频文件 # 将字符串转换为数组，得到一维的short类型的数组 wave_data = np.fromstring(str_data, dtype=np.short) # dtype = s # 赋值的归一化 wave_data = wave_data * 1.0 / (max(abs(wave_data))) # 整合左声道和右声道的数据 # numpy.reshape(a, newshape, order='C') [采样点数, 通道数] 返回值为 *采样点数* 维的数组，每组有 *通道数* 个 wave_data = np.reshape(wave_data, [nframes, 1]) # 最后通过采样点数和取样频率计算出每个取样的时间 time = np.arange(0, nframes) * (1.0 / framerate) # 输出音频信息 feature.textBrowser_3.append("音频信息: " + str(nchannels)) feature.textBrowser_3.append("通道数: " + str(nchannels)) feature.textBrowser_3.append("采样频率: " + str(framerate) + " Hz") feature.textBrowser_3.append("采样点数: " + str(nframes)) feature.textBrowser_3.append("采样时间: " + str(nframes / framerate) + " seconds") # 进度条设置10 feature.progressBar.setValue(10) feature.progressBar_2.setValue(10) # 设置画布 ax = feature.fig1.add_subplot(111) # 调整图像大小 ax.cla() # 画出归一化幅度图像 ax.plot(time, wave_data[:, 0], color='#dc143c') ax.set_title('归一化幅度', fontsize=10) ax.set_xlabel('时间[sec]', fontsize=10) feature.fig1.subplots_adjust(left=0.1, bottom=0.215, right=None, top=None, wspace=None, hspace=None) feature.canvas1.draw() #取一段时间进行分析 # 设置画布 bx = feature.fig8.add_subplot(111) # 调整图像大小 bx.cla() # 画出归一化幅度图像 StartTime0 = np.int(feature.StartTime) DurationTime0 = np.int(feature.DurationTime) StartTime1 = np.int((framerate / 1000) * StartTime0) DurationTime1 = np.int((StartTime1 + DurationTime0 * (framerate / 1000))) # 传入数组的变量必须是整型，而乘除法返回的是浮点，用int()转换 # print(StartTime1) # print(DurationTime1) # 实现20ms的幅度显示 bx.plot(time[StartTime1:DurationTime1], wave_data[StartTime1:DurationTime1, 0], color='#5014DC') bx.set_title('归一化幅度', fontsize=10) bx.set_xlabel('时间[sec]', fontsize=10) feature.fig8.subplots_adjust(left=0.1, bottom=0.215, right=None, top=None, wspace=None, hspace=None) feature.canvas8.draw() # 进度条设置20 feature.progressBar.setValue(20) feature.progressBar_2.setValue(20) # 频域，FFT def Audio_FrequencyDomain(self, feature): # STFT图像绘制 sampling_freq, audio = wavfile.read(feature.path) # 返回采样频率和一维数组，对应8-bit integer PCM编码 T = 20 # 短时傅里叶变换的时长 单位 ms fs = sampling_freq # 采样频率 N = len(audio) # 采样点的个数 audio = audio * 1.0 / (max(abs(audio))) # 归一化 # 计算并绘制STFT的大小 # nperseg ：int, optional # 段长. 默认256 . # f：ndarray # Array of sample frequencies. # t：ndarray # Array of segment times. # Zxx：ndarray # x的STFT. 一般情况下,Zxx 的最后一个轴对应段落时间 . f, t, Zxx = signal.stft(audio, fs, nperseg=np.int(T * fs / 1000)) # 设置进度条30 feature.progressBar.setValue(30) feature.progressBar_2.setValue(30) ax = feature.fig5.add_subplot(111) feature.fig5.subplots_adjust(left=None, bottom=0.215, right=None, top=None, wspace=None, hspace=None) ax.cla() ax.pcolormesh(t, f, np.abs(Zxx), vmin=0, vmax=0.1) # 明显划分出边缘，表现为颜色深浅 ax.set_title('短时傅里叶变换STFT', fontsize=10) ax.set_xlabel('时间 [sec]', fontsize=10) ax.set_ylabel('频率 [Hz]', fontsize=10) feature.canvas5.draw() feature.progressBar.setValue(40) feature.progressBar_2.setValue(40) # 绘制FFT图像 fft_signal = np.fft.fft(audio) fft_signal = abs(fft_signal) # 取绝对值 # 建立频率轴 # 复信号没有负频率，以fs为采样速率的信号，fft的频谱结果是从[0,fs]的。 fft_signal = np.fft.fftshift(fft_signal) # 移动0频率点，画图 # 与上同理 fft_signal = fft_signal[int(fft_signal.shape[0] / 2):] # 频率轴的间隔 freqInteral = (sampling_freq / len(fft_signal)) Freq = np.arange(0, sampling_freq / 2, sampling_freq / (2 * len(fft_signal))) # 设置进度条50 feature.progressBar.setValue(50) feature.progressBar_2.setValue(50) # 计算最高频率 highFreq = (np.argmax(fft_signal[int(len(fft_signal) / 2):len(fft_signal)])) * freqInteral feature.textBrowser_3.append("FFT : 最高频率为: " + str(highFreq)) ax = feature.fig3.add_subplot(111) # 调整图像大小 ax.cla() ax.plot(Freq, fft_signal, color='#dc143c') ax.set_title('FFT 图像', fontsize=10) ax.set_xlabel('频率 [Hz]', fontsize=10) ax.set_ylabel('增益', fontsize=10) feature.fig3.subplots_adjust(left=None, bottom=0.215, right=None, top=None, wspace=None, hspace=None) feature.canvas3.draw() cx = feature.fig10.add_subplot(111) # 调整图像大小 cx.cla() cx.plot(Freq, fft_signal, color='#5014DC') cx.set_title('FFT 图像', fontsize=10) cx.set_xlabel('频率 [Hz]', fontsize=10) cx.set_ylabel('强度 [dB]', fontsize=10) feature.fig10.subplots_adjust(left=None, bottom=0.215, right=None, top=None, wspace=None, hspace=None) feature.canvas10.draw() # 设置进度条60 feature.progressBar.setValue(60) feature.progressBar_2.setValue(60) # 语谱图 def Audio_SpectrogramDomain(self, feature): # 语谱图绘制 f = wave.open(feature.path, "rb") params = f.getparams() nchannels, sampwidth, framerate, nframes = params[:4] strData = f.readframes(nframes) # 读取音频，字符串格式 waveData = np.fromstring(strData, dtype=np.int16) # 将字符串转化为int waveData = waveData * 1.0 / (max(abs(waveData))) # wave幅值归一化 waveData = np.reshape(waveData, [nframes, 1]).T # 合并多声道，.T:数组转置，横坐标为时间 ，变成一维数组，下面会用