Audio Wave Form

Your Words Used Against You

AI startup Lyrebird announced a voice-mimicry algorithm that uses deep learning to create completely fake speech in anyone’s voice using a brief sample. Lyrebird claims their software is “the very first of this kind to allow copying voices in a matter of minutes and control the emotion of the generation.” The ethical and legal implications of this software, which could render every recorded voice sample suspicious, co-opt people’s identities, malign reputations, and profit from others’ talents, are stunning. Fortunately, for the moment, their samples are not quite convincing. —ML