文章目录
- 声压级计算
- 一个文件夹下多个文件的SPL
- 一个文件夹下多组文件的SPL差
- 一个总文件夹下多个子文件夹，每个子文件夹有多组文件，计算spl差。
声压级计算
一个文件夹下多个文件的SPL
执行文件exert
% 功能：遍历data文件夹下的所有文件，计算带A计权的声压级曲线，并在命令行显示SPL（dB）
% 编辑者：Heart_sea
% 日期：2019,4,24
clear;
clc;
close all;
% ======================= preferences set =================================
fs = 12000; % sampling frequency (unit: Hz)
NFFT = fs; % the number of samples used in FFT
pl = 20; % 下限频率
ph = 6000; % 上限频率
N_win = NFFT; % the length of the window，% 规定窗长度是NFFT,暂时这样理解，后面学窗再详细探讨
Window = hamming(N_win); % 或者 WINDOW = NFFT;默认是汉明窗，只给出长度即可
N_overlap = floor(N_win/2); % 重叠是窗长度的一半
A_weighting_flag = 2; % 是否要A计权,0表示不计权，1表示国标的A计权，2表示调用库的A计权
save_image_flag =0; % 是否保存图片,0不存
timecom = 1; % 1 需显示时域曲线，2 无需显示时域曲线
coordinate = 1; % 坐标选择，0表示直角坐标，1表示x轴对数坐标
% ===========计算麦克风的灵敏度
OP_AMP = 5.1; % 麦克风板的放大倍数，采集的是经过放大倍数的数据
MIC_SEN = -38; % 灵敏度级dB
mic_sen = (10^(MIC_SEN./20)).*OP_AMP;%麦克风的灵敏度
% ======================== import data ====================================
maindir = './Data';
File_L1 =dir(maindir); % 获取路径下的所有文件信息
PathNumber=numel(File_L1);% 获取元素的数量
m = 1; % 存放数据列数初始化
switch timecom
case 1
for i = 3:PathNumber
Path=fullfile(maindir,File_L1(i).name);
Path2 = fullfile(File_L1(i).name);
Path2(end-3:end) = [];
note(:,m) = {[File_L1(i,1).name]};
m = m+1;
x = load(Path); % importdata(Path)
x = x./mic_sen;
t = 1/fs:1/fs:length(x)/fs;
plot(t,x)
hold on;
legend(note);
end
title(['时域曲线 '])
xlabel('t (s)')
ylabel('pressure (Pa)')
case 2
end
if timecom == 1
figure;
end
% ======================== 频域 ====================================
m = 1; %列放数据初始化
for i = 3:PathNumber
Path=fullfile(maindir,File_L1(i).name);% 文件夹下文件路径
Path2 = fullfile(File_L1(i).name); % 文件夹下的文件的名字
Path2(end-3:end) = []; %去掉文件的txt
note(:,m) = {[File_L1(i,1).name]}; % 用作legend
m = m+1;
x = load(Path);
x = x./mic_sen;
[Sxx,f] = cpsd(x,x, Window, N_overlap, NFFT, fs);
f = f(pl:ph);
Sxx = Sxx(pl:ph);
% ======================== A-weighting-国标 ===============================
if A_weighting_flag == 1
WeightA = A_curve(f); % get the vaule of weight-A
SPL = 10*log10(Sxx/4e-10)+ WeightA;%各频率上的A计权声压级
SPLA_S = sum(10.^(SPL./10));
SPLA_total = 10*log10(SPLA_S);
% ======================== A-weighting-调用MATLAB库 ===============================
elseif A_weighting_flag == 2
N_f = sum(f<=fs/2);
Weighting = 'A';%Weighting支持‘A’计权或者‘C’计权
% The valid weighting types are: A weighting, C weighting, C-message, ITU-T 0.41, and ITU-R 468–4 weighting.
% Filter class is only applicable for A weighting and C weighting filters.
% https://ww2.mathworks.cn/help/dsp/ref/fdesign.audioweighting.html
Filter_Obj = fdesign.audioweighting('WT,Class',Weighting,1,fs);
H_Filter = design(Filter_Obj);
[Hf,~] = freqz(H_Filter,N_f,fs);
for i = 1:N_f
Sxx(i) = Sxx(i)*((abs(Hf(i)))^2);
end
Sxx_total = sum(Sxx);
SPLA_total = 10*log10(Sxx_total/4e-10);
SPL = 10*log10(Sxx/4e-10);
% ======================== 不计权
elseif A_weighting_flag == 0
Sxx_total = sum(Sxx);
SPLA_total = 10*log10(Sxx_total/4e-10);
SPL = 10*log10(Sxx/4e-10);
end
disp([num2str(Path2) ': ' num2str(SPLA_total) 'dB'])
% ===================== 坐标选择
if coordinate == 0 % 0代表直角坐标
plot(f,SPL)
hold on;
elseif coordinate == 1
semilogx(f,SPL);
set(gca, 'XTick',[20,50,100,200,500,1e3,2e3,6e3] );
% set(gca, 'XTick',[20,50,100,200,500,1e3,2e3,5e3,10e3,20e3] ); % X轴的记号点
set(gca,'xticklabel',get(gca,'xtick'),'yticklabel',get(gca,'ytick'));
hold on;
end
if A_weighting_flag == 0
title(['SPL '])
else
title(['SPLA '])
end
xlabel('f (Hz)')
ylabel('(dB)')
end
legend(note)
if save_image_flag == 1
set(gca,'FontSize',12)
saveas(gcf,'SPL.fig') %可以保存成jpg，fig等不同格式的图片
end
国标计权A_curve
% 功能：根据国标模拟滤波估计计算A计权曲线，单位为dB
% 输入 f：频率
% 输出 y：A计权曲线对应的值
function [y] = A_curve(f)
n1 = 20.6;
n2 = 107.7;
n3 = 737.9;
n4 = 12194;
p = n4^2 .* f.^4;
p1 = n1^2 + f.^2;
p2 = sqrt(n2^2 + f.^2);
p3 = sqrt(n3^2 + f.^2);
p4 = n4^2 + f.^2;
y = 20.*log10(p./(p1 .* p2 .* p3 .* p4))+2;
end
一个文件夹下多组文件的SPL差
执行文件exert
% 功能：遍历data文件夹下的所有文件（off-on），计算带A计权的每组的声压级曲线，并在命令行显示SPL（dB）
% 编辑者：lily
% 日期：2019,4,25
clear;
clc;
close all;
% ======================= preferences set =================================
fs = 12000; % sampling frequency (unit: Hz)
NFFT = fs; % the number of samples used in FFT
fl = 20; % 下限频率
fh = 6000; % 上限频率
point_number = 540000; % the number of sample points
jump_line = 0; % 跳过第jump_line行开始取，0表示从第一行取
micbegin = 1; % 文件序号的第几个开始
micEnd = 4; % 文件序号的第几个结束
N_win = NFFT; % the length of the window，% 规定窗长度是NFFT,暂时这样理解，后面学窗再详细探讨
Window = hamming(N_win); % 或者 WINDOW = NFFT;默认是汉明窗，只给出长度即可
N_overlap = floor(N_win/2); % 重叠是窗长度的一半
A_weighting_flag = 1; % 是否要A计权,0表示不计权，1表示国标的A计权，2表示调用库的A计权
save_image_flag =0; % 是否保存图片,0不存
coordinate = 0; % 坐标选择，0表示直角坐标，1表示x轴对数坐标
smooth = 1; % the switch of SPL smooth, ON:1 OFF:0
% ===========计算麦克风的灵敏度
% 灵敏度与降噪前后的dB值有关，但与降噪前后的dB差值无关
OP_AMP = 5.1; % 麦克风板的放大倍数，采集的是经过放大倍数的数据
MIC_SEN = -38; % 灵敏度级dB
mic_sen = (10^(MIC_SEN./20)).*OP_AMP;%麦克风的灵敏度
% ======================== import data ====================================
figure(1)
for micNumber = micbegin:micEnd
filename = ['err mic' num2str(micNumber) ' - off.txt'];
off = ReadData_Raw(filename,point_number,jump_line);
off = off /mic_sen;%电压/灵敏度=声压
[S_off,f] = cpsd(off,off,Window,N_overlap,NFFT,fs);
filename = ['err mic' num2str(micNumber) ' - on.txt'];
on = ReadData_Raw(filename,point_number,jump_line);
on = on/mic_sen;
[S_on,~] = cpsd(on,on,Window,N_overlap,NFFT,fs);
f = f(fl+1:fh+1);
S_off = S_off(fl+1:fh+1);
S_on = S_on(fl+1:fh+1);
% ======================== A-weighting-国标 ===============================
% 此处关键点：dB相加
if A_weighting_flag == 1
WeightA = A_curve(f); % get the vaule of weight-A
SPL_off = 10*log10(S_off/4e-10)+ WeightA;%各频率上的A计权声压级
SPL_on = 10*log10(S_on/4e-10)+ WeightA;%各频率上的A计权声压级
SPL_off_sum = sum(10.^(SPL_off./10));
SPL_on_sum = sum(10.^(SPL_on./10));
SPL_total_off = 10*log10(SPL_off_sum);
SPL_total_on = 10*log10(SPL_on_sum);
% ======================== A-weighting-调用MATLAB库 ===============================
elseif A_weighting_flag == 2
N_f = sum(f<=fs/2);%返回满足条件的数据个数
Weighting = 'A';%Weighting支持‘A’计权或者‘C’计权
Filter_Obj = fdesign.audioweighting('WT,Class',Weighting,1,fs);
H_Filter = design(Filter_Obj);
[Hf,~] = freqz(H_Filter,N_f,fs);
for i = 1:N_f
S_off(i) = S_off(i)*((abs(Hf(i)))^2);
S_on(i) = S_on(i)*((abs(Hf(i)))^2);
end
Syy_off_sum = sum(S_off);
Syy_on_sum = sum(S_on);
SPL_total_off = 10*log10(Syy_off_sum/4e-10);
SPL_total_on = 10*log10(Syy_on_sum/4e-10);
SPL_off = 10*log10(S_off/4e-10);
SPL_on = 10*log10(S_on/4e-10);
% ======================== 不计权
elseif A_weighting_flag == 0
Syy_off_sum = sum(S_off);
Syy_on_sum = sum(S_on);
SPL_total_off = 10*log10(Syy_off_sum/4e-10);
SPL_total_on = 10*log10(Syy_on_sum/4e-10);
SPL_off = 10*log10(S_off/4e-10);
SPL_on = 10*log10(S_on/4e-10);
end
% ============================= 坐标选择
if coordinate == 0 % 0代表直角坐标
% ======是否需要平滑滤波
if smooth == 0 %不需滤波
subplot (micEnd,1,micNumber)
plot(f,SPL_off);
hold on
plot(f,SPL_on)
grid on
elseif smooth == 1 %平滑滤波
subplot (micEnd,1,micNumber)
% csaps函数可以给定各种边界条件的三次样条插值函数
% csaps（x,y,z）实现光滑拟合，z为权因子，0<z<1,z值越大，越接近真实值，z=0实现线性拟合，z=1实现自然线条
z = 1E-4;
S_off = csaps(f,SPL_off,z);
fnplt(S_off,'b'); %降噪前off
hold on
S_on = csaps(f,SPL_on,z);
fnplt(S_on,'r'); % 降噪后oN
% set(gca,'YTick',-40:10:50);
% set(gca,'XTick',fl:500:fh);
grid on
end
elseif coordinate == 1 % 1代表x对数坐标
% ==========是否需要平滑滤波
if smooth == 0
subplot (micEnd,1,micNumber)
semilogx(f,SPL_off);
hold on
semilogx(f,SPL_on);
set(gca, 'XTick',[20,50,100,200,500,1e3,2e3,6e3] );
% set(gca, 'XTick',[20,50,100,200,500,1e3,2e3,5e3,10e3,20e3] ); % X轴的记号点
set(gca,'xticklabel',get(gca,'xtick'),'yticklabel',get(gca,'ytick'));
grid on
elseif smooth == 1 %平滑滤波
subplot (micEnd,1,micNumber)
% csaps函数可以给定各种边界条件的三次样条插值函数
% csaps（x,y,z）实现光滑拟合，z为权因子，0<z<1,z值越大，越接近真实值，z=0实现线性拟合，z=1实现自然线条
z = 1E-4;
S_off = csaps(f,SPL_off,z);
semilogx(f,([S_off.coefs(:,end);S_off.coefs(end,end)]),'b'); % 绘制样条函数图形
hold on
S_on = csaps(f,SPL_on,z);
semilogx(f,([S_on.coefs(:,end);S_on.coefs(end,end)]),'r');
set(gca, 'XTick',[20,50,100,200,500,1e3,2e3,6e3] );
set(gca,'xticklabel',get(gca,'xtick'),'yticklabel',get(gca,'ytick'));
grid on
end
end
k = SPL_total_off - SPL_total_on;%s是off，c是on
disp(['err' num2str(micNumber) ' = ' num2str(k) 'dB'] );
disp(['第' num2str(micNumber) '组SPL：降噪前=' num2str(SPL_total_off) 'dB' ' 降噪后=' num2str(SPL_total_on) 'dB']);
% ============= 标题
if A_weighting_flag == 0
title(['SPL ' num2str(micNumber)])
else
title(['SPLA ' num2str(micNumber)])
end
xlabel('f (Hz)')
ylabel('(dB)')
end
% =============保存图片
if save_image_flag == 1
set(gca,'FontSize',12)
saveas(gcf,'ANCEffectFig.fig') %可以保存成jpg，fig等不同格式的图片
end
一个总文件夹下多个子文件夹，每个子文件夹有多组文件，计算spl差。
执行文件exert
## 一个总文件夹下多个子文件夹，每个子文件夹有多组文件
### 执行文件exert
```java
% 功能：遍历data文件夹下的所有子文件夹下的文件（off-on），计算带A计权的每组的声压级曲线，并在命令行显示SPL（dB）
% 编辑者：lily
% 日期：2019,4,25
clear;
clc;
close all;
% ======================= preferences set =================================
maindir = './Data';
fs = 12000; % sampling frequency (unit: Hz)
NFFT = fs; % the number of samples used in FFT
fl = 20; % 下限频率
fh = 6000; % 上限频率
max = 4; % 子文件夹最多的文件个数
point_number = 80000; % the number of sample points
jump_line = 0; % 跳过第jump_line行开始取，0表示从第一行取
N_win = NFFT; % the length of the window，% 规定窗长度是NFFT,暂时这样理解，后面学窗再详细探讨
Window = hamming(N_win); % 或者 WINDOW = NFFT;默认是汉明窗，只给出长度即可
N_overlap = floor(N_win/2); % 重叠是窗长度的一半
A_weighting_flag = 1; % 是否要A计权,0表示不计权，1表示国标的A计权，2表示调用库的A计权
save_image_flag =0; % 是否保存图片,0不存
coordinate = 1; % 坐标选择，0表示直角坐标，1表示x轴对数坐标
smooth = 1; % the switch of SPL smooth, ON:1 OFF:0
% ========计算麦克风的灵敏度
% 灵敏度与降噪前后的dB值有关，但与降噪前后的dB差值无关
OP_AMP = 5.1; % 麦克风板的放大倍数，采集的是经过放大倍数的数据
MIC_SEN = -38; % 灵敏度级dB
mic_sen = (10^(MIC_SEN./20)).*OP_AMP;%麦克风的灵敏度
% ======================== 循环 import data ===============================
[Note_legend,~] = importdata(maindir,fs,NFFT,fl,fh,max,point_number,jump_line,...
N_win,Window,N_overlap,A_weighting_flag,coordinate,smooth,mic_sen);
% =============保存图片
if save_image_flag == 1
set(gca,'FontSize',12)
saveas(gcf,'ANCEffectFig.fig') %可以保存成jpg，fig等不同格式的图片
end
调用函数importdata
function [Note_legend,Note_disp] = importdata(maindir,fs,NFFT,fl,fh,max,...
point_number,jump_line,~,Window,N_overlap,A_weighting_flag,coordinate,smooth,mic_sen)
File_L1 =dir(maindir); % 获取data里的内容
PathNumber=numel(File_L1);% 获取结构体的行数
for i = 3:PathNumber %读取数据从第三行到最后一行
figini = 2;%fig初始化,3-2=1,每一组都要初始化
Path = fullfile(maindir,File_L1(i).name);%前两阶文件夹的名称，字符串格式
% ==============设置lengend
Path2 = fullfile(File_L1(i).name);%获取第二层文件夹的名字，字符串格式
dat = dir(Path);%获取第二层文件夹的所有txt文件，struct结构体格式
row = 1;
for j = 3 : length( dat ) %循环dat名称
str = strcat(Path2,fullfile(dat( j ).name)); % str获取文件夹及其文件名字，并连接
str(end-3:end) = []; % 将字符串的后四个字符串置成空的，去掉.txt
Note_legend(row,i-2) = {str};
row = row+1;
end
% ======================== 循环dat数据（子文件夹里的文件） ================
colum =1; % 存放数据到列初始化
v = 2; % subplot第几个fig初始化
for m = 3:2:length( dat )
% ==============设置disp
str1 = strcat(Path2,fullfile(dat( m ).name));
str1(end-8:end) = [];
Note_disp (:,colum) = {str1 };
colum = colum+1;
% ==============开始导入数据
% off = load(fullfile(Path,dat(m).name));
filename = fullfile(Path,dat(m).name);
off = ReadData_Raw(filename,point_number,jump_line);
off = off /mic_sen;%电压/灵敏度=声压
[S_off,f] = cpsd(off,off,Window,N_overlap,NFFT,fs);
filename = fullfile(Path,dat(m+1).name);
on = ReadData_Raw(filename,point_number,jump_line);
on = on /mic_sen;
[S_on,~] = cpsd(on,on,Window,N_overlap,NFFT,fs);
f = f(fl+1:fh+1);
S_off = S_off(fl+1:fh+1);
S_on = S_on(fl+1:fh+1);
% ======================== A-weighting-国标 ===============================
% 此处关键点：dB相加
if A_weighting_flag == 1
WeightA = A_curve(f); % get the vaule of weight-A
SPL_off = 10*log10(S_off/4e-10)+ WeightA;%各频率上的A计权声压级
SPL_on = 10*log10(S_on/4e-10)+ WeightA;%各频率上的A计权声压级
SPL_off_sum = sum(10.^(SPL_off./10));
SPL_on_sum = sum(10.^(SPL_on./10));
SPL_total_off = 10*log10(SPL_off_sum);
SPL_total_on = 10*log10(SPL_on_sum);
% ======================== A-weighting-调用MATLAB库 ===============================
elseif A_weighting_flag == 2
N_f = sum(f<=fs/2);%返回满足条件的数据个数
Weighting = 'A';%Weighting支持‘A’计权或者‘C’计权
Filter_Obj = fdesign.audioweighting('WT,Class',Weighting,1,fs);
H_Filter = design(Filter_Obj);
[Hf,~] = freqz(H_Filter,N_f,fs);
for i = 1:N_f
S_off(i) = S_off(i)*((abs(Hf(i)))^2);
S_on(i) = S_on(i)*((abs(Hf(i)))^2);
end
Syy_off_sum = sum(S_off);
Syy_on_sum = sum(S_on);
SPL_total_off = 10*log10(Syy_off_sum/4e-10);
SPL_total_on = 10*log10(Syy_on_sum/4e-10);
SPL_off = 10*log10(S_off/4e-10);
SPL_on = 10*log10(S_on/4e-10);
% ======================== 不计权
elseif A_weighting_flag == 0
Syy_off_sum = sum(S_off);
Syy_on_sum = sum(S_on);
SPL_total_off = 10*log10(Syy_off_sum/4e-10);
SPL_total_on = 10*log10(Syy_on_sum/4e-10);
SPL_off = 10*log10(S_off/4e-10);
SPL_on = 10*log10(S_on/4e-10);
end
% ============================= 坐标选择
fig = m-v;% fig的第几个。m=3时，fig=1，m=5时，fig=2,依次类推
if coordinate == 0 % 0代表直角坐标
if smooth == 0 %不需滤波
subplot (max,1,fig)
plot(f,SPL_off);
hold on
plot(f,SPL_on)
grid on
elseif smooth == 1 %平滑滤波
subplot (max,1,fig)
% csaps函数可以给定各种边界条件的三次样条插值函数
% csaps（x,y,z）实现光滑拟合，z为权因子，0<z<1,z值越大，越接近真实值，z=0实现线性拟合，z=1实现自然线条
z = 1E-4;
S_off = csaps(f,SPL_off,z);
fnplt(S_off,''); %降噪前off
hold on
S_on = csaps(f,SPL_on,z);
fnplt(S_on,''); % 降噪后oN
% set(gca,'YTick',-40:10:50);
% set(gca,'XTick',fl:500:fh);
grid on
end
elseif coordinate == 1 % 1代表x对数坐标
if smooth == 0
subplot (max,1,fig)
semilogx(f,SPL_off);
hold on
semilogx(f,SPL_on);
set(gca, 'XTick',[20,50,100,200,500,1e3,2e3,6e3] );
% set(gca, 'XTick',[20,50,100,200,500,1e3,2e3,5e3,10e3,20e3] ); % X轴的记号点
set(gca,'xticklabel',get(gca,'xtick'),'yticklabel',get(gca,'ytick'));
axis([fl,fh,-inf,inf]);
grid on
elseif smooth == 1 %平滑滤波
subplot (max,1,fig)
% csaps函数可以给定各种边界条件的三次样条插值函数
% csaps（x,y,z）实现光滑拟合，z为权因子，0<z<1,z值越大，越接近真实值，z=0实现线性拟合，z=1实现自然线条
z = 1E-4;
S_off = csaps(f,SPL_off,z);
semilogx(f,([S_off.coefs(:,end);S_off.coefs(end,end)]),''); % 绘制样条函数图形
hold on
S_on = csaps(f,SPL_on,z);
semilogx(f,([S_on.coefs(:,end);S_on.coefs(end,end)]),'');
set(gca, 'XTick',[20,50,100,200,500,1e3,2e3,6e3] );
set(gca,'xticklabel',get(gca,'xtick'),'yticklabel',get(gca,'ytick'));
axis([fl,fh,-inf,inf]);
grid on
end
end
legend(char(Note_legend(m-2:m-2+1,:)))
v = v+1;
k = SPL_total_off - SPL_total_on;%s是off，c是on
disp([num2str(Note_disp{fig}) ':' num2str(k) 'dB'])
% ============= 标题
if A_weighting_flag == 0
title(['SPL ' num2str(fig)])
else
title(['SPLA ' num2str(fig)])
end
xlabel('f (Hz)')
ylabel('(dB)')
end
end
end
