Digital signal processing spectrum inversion for analog speech channels
Masahichi Kishi, Jun Tajima, Noboru Kanmuri, Seizo Seki · IEEE Transactions on Vehicular Technology · 1983
Speech security communication systems applied to radio analog speech channels become more and more desirable as radio channel capacity increases and service area spreads. Analog spectrum inversion used in existing radio communication systems is suitable for preventing eavesdropping and ensuring security, but it suffers from inevitable speech quality degradations. It is shown that digital spectrum inversion can be accomplished merely by alternating the sign of the sampled data to avoid distortion. Digital inversion is theoretically deduced by the discrete Fourier transform and verified by experiments. The aperture effect is also discussed from the viewpoint of spectrum inversion. Digital spectrum inversion is implemented with exclusive OR gates and a one digit binary counter, which is easily integrated onto a large scale integration (LSI) chip to avoid element value deviation. Signal-to-noise ratio (SNR) and distortion are observed to be more than 50 dB and -50 dB right across the 0.3-3.7 kHz frequency band, when an 8 kHz sampling rate is adopted for digital processing.