Hi,
What type of problem do you see calculating the FFT? do you get any results at all, or an error returning from the function?
I'm not very familir with "time response of an octave" but you could also try to filter the band (octave)that you are interested in and calculate RMS value of the filtered data over different periods of time and get a time response.
Just my 2 cents; keep me posted.
Regards,
Juan Carlos
N.I.