FM audio is part of the analog TV signal that carries sound for a video program. The FM audio band for each channel in terrestrial TV is located at a fixed offset from the video band. In a FM modulated audio signal, the carrier frequency is varied proportional to the input audio signal, and the magnitude of the signal is constant.
A basic FM audio demodulator is shown in Fig.1. The RF FM audio signal is first down-converted to the baseband. The baseband FM signal is filtered in the main filter to attenuate adjacent channels and other out of band signals. Then, the CORDIC technique is applied to do the FM demodulation. The CORDIC technique allows the calculation of the complex signal phase and magnitude from its real and imaginary components. Since CORDIC does phase estimation, rather than frequency estimation, the complex signal has to be modified as follows before going to the CORDIC phase detector:
s’(t) = s(t)∙conj(s(t–ts)),
here s(t) is the baseband complex FM audio signal, s’(t) is the output signal, and ts is the sampling period. This transformation produces the signal s’(t) that has a constant magnitude and a phase that is equal to the phase difference between the input signal and the delayed input signal, i.e. the frequency of the input signal. Therefore, the CORDIC phase estimate from s’(t) is proportional to the frequency of s(t).
The output of CORDIC goes through a baseband filter that removes frequencies outside of the audio range. It is also an FM de-emphasis filter.
For Stereo FM, the baseband signal after the FM demodulator contains other subcarriers: the left and right channel difference, pilot and RDS/RBDS signal. Additional post-processing is done to recover these signals.