Divide .wav file into equal segments
古いコメントを表示
I've been having some trouble dividing my 159567 ms audio file into 3000 ms segments. I've seen on previous threads about using the 'bufer' function, but I'm not quite sure how to implement it here. What ends up happening here is that the correct number of segments are created, but the audio file doesn't get cut into its 3000 ms chunks (instead I have 53 audio files that are exactly the same). Any help would be appreciated.
%epoch into 3 sec nonoverlapping segments
for i = 1
[y, Fs] = audioread('Myfile.wav');
stims{i} = y;
stiml(i) = length(stims{i});
stiml(i) = stiml(i)/44.1; %put lengths into milliseconds in time (not samples). Length: 2 min, 39 s
end
t1 = 1;
t2 = 3000; %ms
k = 1;
while t2 <= stiml %equal to 159567 ms
new_seg = [t1,t2];
File = sprintf('Part%02d.wav', k);
audiowrite(File, y, Fs);
t1=t1+3000;
t2=t2+3000;
k=k+1;
end
2 件のコメント
Walter Roberson
2020 年 5 月 27 日
You are proceeding by 3000 samples, not by 3000 ms. To get 3000 milliseconds in terms of samples you need to multiply Fs by 3000 (milliseconds) / 1000 (milliseconds per second)
Christine
2020 年 5 月 27 日
採用された回答
Walter Roberson
2020 年 5 月 28 日
編集済み: Walter Roberson
2020 年 5 月 28 日
The below will split all .wav in a given directory into 3 second buffers. A directory named after the file will be created inside outdir for each file, and the split files will be named by Part inside the relevant directory.
This code does not assume that the sample frequency is the same for all files.
This code does not assume mono source, and does not assume that the number of channels is the same for each file.
%epoch into 3 sec nonoverlapping segments
projectdir = 'location of files to split';
outdir = 'location of directory to split into'; %can be the same as projectdir
dinfo = dir( fullfile(projectdir, '*.wav') );
filenames = fullfile( projectdir, {dinfo.name} );
nfiles = length(dinfo);
stims = cell(nfiles,1);
stiml = zeros(nfiles,1);
Fss = zeros(nfiles,1);
for K = 1 : nfiles
[y, Fss(K)] = audioread(filenames{K});
stims{K} = y;
stiml(K) = size(y,1) / Fss(K) * 1000;
end
for K = 1 : nfiles
[~, basename, ext] = fileparts( filenames{K} );
thisoutdir = fullfile(outdir, basename);
if ~exist(thisoutdir, 'dir');
try
mkdir(thisoutdir);
catch ME
fprintf('count not make the output directory "%s", skipping file\n', thisoutdir);
continue;
end
end
nchan = size(stims{K},2);
Fs = Fss(K);
buffered_chans = arrayfun(@(Cidx) buffer(stims{K}(:,Cidx), floor(3*Fss{K})), 1:nchan, 'uniform', 0);
nparts = size(buffered_chans{1},2);
for part = 1 : nparts
File = fullfile(thisoutdir, sprintf('Part%02d.wav', part));
this_segment = cell2mat(cellfun(@(C) C(:,part), buffered_chans, 'uniform', 0));
audiowrite(File, this_segment, Fs);
end
end
5 件のコメント
Christine
2020 年 5 月 28 日
Walter Roberson
2020 年 5 月 28 日
編集済み: Walter Roberson
2020 年 5 月 28 日
The first C was a typing mistake on my part.
The Cidx is a dummy parameter name for the purposes of the anonymous function. It is acting as a channel index.
What is going on there is that buffer() takes a vector of values and re-arranges it into a 2D array, each column being a segment. But any given input might have more than one channel, so you have to process each channel separately through buffer. Channels in audio are columns, so 1:nchan is a list of the channel numbers for this particular input. The arrayfun() is therefore acting to iterate over the channels.
The second C is a dummy parameter name for the purposes of the second anonymous function. It is acting to hold one of the channels that has been put through buffer(). Splitting into channels in the first function created a cell array element for each channel, and the cellfun() is then acting on each of those blocks in turn. In the context of the for loop, it is extracting one column from the buffer()'d values, which is to say it is extracting one segment at a time. It does that for each of the channels, and then the cell2mat() puts those back together into a multi-channel segment ready to be written.
Christine
2020 年 5 月 28 日
awais khan
2021 年 6 月 29 日
Brace indexing is not supported for variables of this type.
I got this error.
You are right, it needed a minor fix.
%epoch into 3 sec nonoverlapping segments
projectdir = fullfile(matlabroot, 'toolbox', 'audio', 'samples');
outdir = 'junk_out'; %can be the same as projectdir
dinfo = dir( fullfile(projectdir, '*.wav') );
filenames = fullfile( projectdir, {dinfo.name} );
nfiles = length(dinfo);
stims = cell(nfiles,1);
stiml = zeros(nfiles,1);
Fss = zeros(nfiles,1);
for K = 1 : nfiles
[y, Fss(K)] = audioread(filenames{K});
stims{K} = y;
stiml(K) = size(y,1) / Fss(K) * 1000;
end
for K = 1 : nfiles
[~, basename, ext] = fileparts( filenames{K} );
thisoutdir = fullfile(outdir, basename);
if ~exist(thisoutdir, 'dir');
try
mkdir(thisoutdir);
catch ME
fprintf('count not make the output directory "%s", skipping file\n', thisoutdir);
continue;
end
end
nchan = size(stims{K},2);
Fs = Fss(K);
buffered_chans = arrayfun(@(Cidx) buffer(stims{K}(:,Cidx), floor(3*Fss(K))), 1:nchan, 'uniform', 0);
nparts = size(buffered_chans{1},2);
for part = 1 : nparts
File = fullfile(thisoutdir, sprintf('Part%02d.wav', part));
this_segment = cell2mat(cellfun(@(C) C(:,part), buffered_chans, 'uniform', 0));
audiowrite(File, this_segment, Fs);
fprintf('wrote "%s"\n', File)
end
end
その他の回答 (0 件)
カテゴリ
ヘルプ センター および File Exchange で Audio I/O and Waveform Generation についてさらに検索
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
Translated by 
