Our frequency domain approach for pitch shifting will use the “phase vocoder” method, that effectively is an FFT/inverse FFT. The term “vocoder” is somewhat misleading here, because it has nothing to do with a real “vocoder”.
FFT will introduce latency too, but the buffers can be kept relatively small and the input can be shifted by exactly this fixed amount of latency for the final mix internally.
Complex mathematical formulas will be used to extract the frequency components from the signal and their corresponding magnitudes. After the frequency is known, it can be modified in controlled manner, thus making pitch and frequency shifting possible in realtime.
If the frequency was modified across all magnitudes equally, the inverse FFT is applied, leaving the audio with a changed pitch. Like the time domain approach of JAX ESSENTIAL SERIES P2, this will produce “chipmonkeyfication” effects, because all transients and formants will be shifted this way too, proportional with the fundamental frequency.
The frequency domain approach has some side effects too. The audio will be smeared slightly but equally across the entire range. Grainy results are less audible than with the time domain approach. It is not well situated for using with strong transient audio material. Excellent results can be achieved with ambient audio material.