# Voice Cloning

> AI copies a person's voice from a short recording. Then it can say new words in it.

**Field:** Artificial Intelligence  
**Category:** Multimodal  
**Also known as:** AI Voice Cloning, Voice Replication, Synthetic Voice

## Think of it like

It is like a parrot that learns to sound just like you. Then it can say new things in your voice.

## Definition

Voice cloning is AI that copies one person's voice. New text can then be spoken in it. The model learns how the voice sounds from a recording, sometimes less than a minute long. After that, any text can be read out in that voice, often in other languages too. It is used for dubbing, for people losing their voice to illness, and to keep one narrator across many recordings. Scammers use it too, so honest services ask for proof of consent.

## Example

Jack is losing his voice to an illness. While he still can, he records himself reading for half an hour. A cloned voice then lets him keep talking to his family in his own voice.

## Why it matters

It can give a voice back to people who lose theirs. It can also fool a parent with a fake call from their child.

## Common mistake

Trusting a voice on the phone because it sounds right. For money or secrets, check with a second way, such as a call back to a known number.

## Three principles

- **Short recording.** Learns the voice.
- **Any new text.** Same voice.
- **Consent matters.** Scams use it too.

## Related terms

- [Text-to-Speech](https://learn.benibix.com/term/ai/text-to-speech)
- [AI Lip Sync](https://learn.benibix.com/term/ai/ai-lip-sync)
- [Generative AI](https://learn.benibix.com/term/ai/generative-ai)
- [Deepfake](https://learn.benibix.com/term/ai/deepfake)
- [AI Voice Scam](https://learn.benibix.com/term/ai/ai-voice-scam)

## Told apart from

- [Text-to-Speech vs Voice Cloning](https://learn.benibix.com/compare/text-to-speech-vs-voice-cloning): People often think voice cloning is just text-to-speech with a nicer voice. Text-to-speech reads text in one of a set of stock voices that the tool provides. Voice cloning first copies one real person's voice from a recording, then reads any text in that voice. That is why cloning needs consent, and plain text-to-speech does not.

## Taught in

- [Runway and ElevenLabs](https://learn.benibix.com/learn/ai-ai-tools/runway-and-elevenlabs): AI Apps and Tools, 10/10

![A short recording of one person's voice is turned into a voice profile. New text is then read out in that same voice.](https://learn.benibix.com/diagrams/mm-voice-cloning.svg)

---

Source: https://learn.benibix.com/term/ai/voice-cloning
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-17
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
