Echo is experienced in normal PSTN connections and also in VoIP connections, but due to the inherent delay in VoIP, it can actually be more noticeable. Echo that a caller hears is probably the result of the far-end, where an impedance mismatch bounces signals back from where they came, and in today’s global telecoms networks calls can transit a mixture of IP and PSTN network, on copper, fiber-optic, and wireless transmission legs. Due to notable delays, there are echoes of one's own words audibly discernable.
Research indicates that when there is a delay of 35ms or less in one-way transmission (round-trip delay of <70ms), the speaker cannot distinguish the echo from an acceptable level of sidetone. However, as one-way delay increases beyond 35ms, the echo grows more problematic. The longer the delay, the more dramatic and invasive the echo becomes.