If I understand correctly, your data was sampled at different rates, so you need the number of samples in temperature (1921) to match the number of samples in frequency (1200).
We can do this by firstly sampling temperature at a higher rate, so that you have 2400 temperature points (2400 is an integer multiple of 1200). And then sampling the 2400 temperature points at a lower rate (i.e. every second point), so that you have exactly 1200 points.
First step: Resample the temperature data at a higher rate using MATLAB's interp function (see the MATLAB documentation) so that you have 2400 temperature points.
Second step: Remove every second sample to get a temperature vector of size 1200: temperature_size1200 = temperature_size2400(1:2:end);
Alternatively, a rough way to do it (but less precise), is to just resample temperature at the lower rate: temperature_size1200 = temperature_size1921(1:(1921/1200):end);
The problem with this approach, is that (1921/1200) is not an integer, so you lose some accuracy than before if you interpolate the data first at an integer multiple of 1200.