Voice Cloning

Voice cloning is the synthesis of a target speaker's voice from a short reference audio sample, allowing a TTS system to produce new speech in that speaker's timbre, accent, and (to a lesser extent) speaking style. Modern systems from vendors like ElevenLabs and Cartesia can produce a usable clone from seconds to minutes of clean reference audio, rather than the hours required by older speaker-adaptation approaches. The capability raises real consent and authentication concerns: deployments that clone real people's voices need explicit permission from the speaker, and downstream consumers (banks, support lines) increasingly cannot treat a familiar voice as proof of identity.