Abstract: A double-talk detector (109) for an acoustic echo canceler (104) of a VoIP terminal (101) converts (216,226) received far-end signals (212) and transmitted near-end signals (222) into the frequency domain and high-pass filters out (218,228) low-frequencies representing noise to obtain a complex reference signal and a complex error signal, respectively. It then correlates (230) the complex signals and computes (232) the instantaneous error energy (234) and smoothed (long-term average) error energy (235) of the complex error signal. If the convergence value is below 0.06 (300), the signals are converged and double-talk detection is enabled (306). If the convergence value is above 0.09 (312), the signals are diverged and double-talk detection is disabled (316). If double-talk detection is not enabled, an error threshold (236) is set (304,318) to the value of the smoothed error energy.

Abstract: A double-talk detector (109) for an acoustic echo canceler (104) of a VoIP terminal (101) converts (216,226) received far-end signals (212) and transmitted near-end signals (222) into the frequency domain and high-pass filters out (218,228) low-frequencies representing noise to obtain a complex reference signal and a complex error signal, respectively. It then correlates (230) the complex signals and computes (232) the instantaneous error energy (234) and smoothed (long-term average) error energy (235) of the complex error signal. If the convergence value is below 0.06 (300), the signals are converged and double-talk detection is enabled (306). If the convergence value is above 0.09 (312), the signals are diverged and double-talk detection is disabled (316). If double-talk detection is not enabled, an error threshold (236) is set (304,318) to the value of the smoothed error energy.

Abstract: A voice-activity detector (VAD 104) takes (214) a currently-received set and a previously-received set of samples of a time-domain (voice) signal, converts (216) them into a frequency-domain representation of the signal, filters out (218) negative and low (noise) frequencies, weights (220) the energies of frequency bins (ranges) of the remaining frequencies proportionately to their frequencies, and computes (220) the total power of the ranges. It first initializes (226) by determining (304, 306) if power peaks of any of the ranges exceed a first threshold (ceiling 228); if not, it lowers (302) the ceiling and continues initializing, and if so, it ends initializing (308), indicates (334) that voice has been detected, sets (330) the ceiling to the highest peak, and stores (332) the total power as a “smoothed” power.