>>106475004
Two reasons: firstly higher pitched sounds are always harder to block (don't ask me why, but something about the higher frequences makes them much harder to match exactly), and rapidly changing sounds are also much harder to block - since the active mechanism has to itself process the sound, generate an inverse soundwave and then generate it from the speakers, there's inherent lag in this. Works perfectly fine for uniform drones (like engine noise, to some degree background crowd chatter or engine noise), speech is a constantly varying sound and even a tiny bit of lag would result in the inverse waveform always lagging behind the real sound and therefore never quite cancelling it properly.