ElevenLabs offers an AI voice cloning tool that lets users create a digital replica of any voice by uploading audio samples, which are analyzed to capture vocal characteristics like tone, pitch, accent, and speaking style. The cloned voice can then generate new speech from text.

The platform offers two tiers of voice cloning. Instant Voice Cloning creates a voice clone from short audio samples (around 1 to 5 minutes) near-instantly, without training a custom model; it is available to paid Starter plan subscribers and above. Professional Voice Cloning trains a dedicated voice model using at least 30 minutes of audio, ideally 2 or more hours, and produces results the company describes as virtually indistinguishable from the original speaker; it requires a Creator plan or higher and voice-captcha verification to confirm the user is cloning their own voice. Both tiers use deep learning to analyze vocal characteristics. ElevenLabs has also implemented a 'No-Go Voices' policy that blocks cloning of political figures and other high-risk voices.