You can just start using the data as if it were the higher sampling rate. However, by doing so you raise the frequency of the samples; in the case of audio, the effect would as if the original source had been one octave higher than it really was. It is valid, for example, to do
[samples, Fs] = audioread('YourFile.wav');
sound(samples, 2*Fs) %will play at double the original pitch
Somehow I suspect that is not your intention. I suspect that for your purpose the answer is NO.
Consider: 25000 samples at 12500 samples per second is 2 seconds. If you double the sampling rate without changing the number of samples, then you would be wanting 25000 samples/s and you would still have 25000 samples. Clearly 25000 samples at 25000 samples/s can only take 1 seconds, not the 2 seconds of the original. If you wanted 2 seconds of output at 25000 samples per second, you would need 50000 samples, not the original 25000 samples.
