Updates to adding sounds together

jp_addnoise.m

@@ -30,7 +30,7 @@ function jp_addnoise(soundfiles, cfg)
 %
 %  jp_addnoise(inDir, cfg);
 %
-%  From https://github.com/jpeelle/jp_matlab
+% From https://github.com/jpeelle/jp_matlab
 
 
 if ~isfield(cfg, 'prestim') || isempty(cfg.prestim)

jp_addsounds.m

@@ -1,5 +1,5 @@
 function y3 = jp_addsounds(y1, y2, fs, cfg)
-% Add sounds
+%JP_ADDSOUNDS Add  two sounds (one target, one distractor)
 %
 % y1 target
 % y2 distractor

jp_crossfade.m

@@ -0,0 +1,24 @@
+function [y, fs] = jp_crossfade(y1, y2, fs, rampSec)
+%JP_CROSSFADE Very basic crossfade for two sounds.
+%
+% JP_CROSSFADE(Y1, Y2, FS) fades sound Y1 to sound Y2 using a linear ramp.
+%
+% JP_CROSSFADE(Y1, Y2, FS, RAMPSEC) uses a specified ramp duration (default
+% is 0.1 seconds).
+%
+% From https://github.com/jpeelle/jp_matlab
+
+if nargin < 4 || isempty(rampSec)
+    rampSec = 0.1;
+end
+
+rampSamples = rampSec*fs;
+fadeIn = [1:rampSamples]'/rampSamples;
+fadeOut = flipud(fadeIn);
+
+y1(end-rampSamples+1:end) = y1(end-rampSamples+1:end) .* fadeOut;
+y2(1:rampSamples) = y2(1:rampSamples) .* fadeIn;
+
+y = zeros(length(y1)+length(y2)-rampSamples, 1);
+y(1:length(y1)) = y1;
+y(length(y1)+1-rampSamples:end) = y(length(y1)+1-rampSamples:end) + y2;

jp_mag2db.m

@@ -0,0 +1,3 @@
+function dB = jp_mag2db(y)
+%JP_MAG2DB Convert magnitude (e.g., RMS) to dB
+dB = 20*log10(y);

jp_stringtogethersounds.m

@@ -0,0 +1,110 @@
+function [y, fs] = jp_stringtogethersounds(files, Cfg)
+%JP_STRINGTOGETHERSOUNDS Make one long sound file from shorter ones.
+%
+% JP_STRINGTOGETHERSOUNDS(FILES) takes individual sound files and puts
+% them togehter into a long sound (e.g., for presenting during an
+% experiment). FILES is assumed to be a cell array of sound file names (all
+% with the same sampling rate).
+%
+% Cfg has the following fields to set options:
+%
+%  pauseBetweenSec   Pause between the end of one sound and the beginning of the next (sec) (default 2)
+%  noiseFile         If not empty, add consistent noise throughout (will concatenate if too short)
+%  SNR               SNR to use for noise
+%  noisePreSoundSec  How long noise plays before sounds start
+%  noisePostSoundSec How long noise plays after sounds stop
+%
+%
+% From https://github.com/jpeelle/jp_matlab
+
+if nargin < 2
+    Cfg = [];
+end
+
+if ~isfield(Cfg, 'pauseBetweenSec') || isempty(Cfg.pauseBetweenSec)
+    Cfg.pauseBetweenSec = 2;
+end
+
+if ~isfield(Cfg, 'noiseFile')
+    Cfg.noiseFile = '';
+end
+
+if ~isfield(Cfg, 'noisePreSoundSec') || isempty(Cfg.noisePreSoundSec)
+    Cfg.noisePreSoundSec = 1;
+end
+
+
+if ~isfield(Cfg, 'noisePostSoundSec') || isempty(Cfg.noisePostSoundSec)
+    Cfg.noisePostSoundSec = 1;
+end
+
+% Read in the first file just to get the sampling rate to make sure this is
+% consistent across everything
+[y, fs] = audioread(files{1});
+
+% If we are using noise, read it in
+if ~isempty(Cfg.noiseFile)
+    assert(exist(Cfg.noiseFile, 'file')>0, 'Specified noise file %s not found.', Cfg.noiseFile);
+    [yNoise, fsNoise] = audioread(Cfg.noiseFile);
+    assert(fsNoise==fs, 'Sampling rate of the noise (%i) does not match the sounds (%i).', fsNoise, fs);
+end
+
+% Read in sounds but don't do anything
+for fileInd = 1:length(files)
+    thisFile = files{fileInd};
+    [ySounds{fileInd}, fsSound] = audioread(thisFile);
+    assert(fsSound==fs, 'Sampling rate of sound %s (%i) does not match that of the first sound (%i).', thisFile, fsSound, fs);    
+end
+
+
+% String together the sounds
+pauseSamp = zeros(Cfg.pauseBetweenSec * fs, 1);
+
+y = [];
+for soundInd = 1:length(ySounds)
+    y = [y; ySounds{soundInd}; pauseSamp];
+end
+
+% If we are adding noise, do that
+if ~isempty(Cfg.noiseFile)
+    % Make sure the noise is long enough. If not, repeat it.
+    while length(yNoise) < length(y)
+        warning('The noise file was not long enough to cover all the sounds; repeating it with a little bit of crossfade.');
+        yNoise = jp_crossfade(yNoise, yNoise, fs, 0.1);
+    end
+
+    % If the noise is TOO long, trim it. (Doing so now helps make sure that
+    % the SNR calculations only take into account the important parts of
+    % the noise.
+    noisePreSoundSamples = Cfg.noisePreSoundSec*fs;
+    noisePostSoundSamples = Cfg.noisePostSoundSec*fs;
+    totalLengthSamples = length(y) + noisePreSoundSamples + noisePostSoundSamples;
+    yNoise = yNoise(1:totalLengthSamples);
+
+    % get level for noise file
+    rmsNoise = jp_rms(yNoise);
+    dbNoise = jp_mag2db(rmsNoise);
+
+    % get level for sounds
+    dbSignal = zeros(1,length(ySounds));
+    for soundInd = 1:length(ySounds)
+        dbSignal(soundInd) = jp_mag2db(jp_rms(ySounds{soundInd}));
+    end
+
+    dbSignalAvg = mean(dbSignal);
+
+    % Scale noise to get right SNR
+    targetDb = dbSignalAvg - Cfg.SNR; % target for noise dB
+    targetRMS = 10^(targetDb/20);
+    scaleFactor = targetRMS/rmsNoise;    
+    yNoise = yNoise * scaleFactor;
+
+    y = [zeros(noisePreSoundSamples,1); y; zeros(noisePostSoundSamples,1)] + yNoise;    
+end
+
+
+end % main function
+
+% rms function
+
+