This is fairly easy at first. You just need to pad the signal a2 with enough zeros to equal 10 seconds. This requires knowledge of the signals sampling fequency.
fs = % your audio fs
timeStart = 10; % time in seconds
a2Pad = [zeros(1,fs*timeStart),a2]; % if a2 is vertical use [zeros(fs*timeStart,1);a2];
newaudio = a1+a2pad
However, this will likely result in the arrays being different sizes so you need to pad the end of the shorter one ( or cut the longer) to be able to add them. This can be done in nearly the same way.
