AIWiki
Malaysia
Back to all articles
Applicationsvoice cloningtext-to-speechdeepfake audio

Voice Cloning

4 min readUpdated August 2026
Voice Cloning
Type
Generative AI audio technology
Description
AI creation of a digital replica of a person's voice
Key providers
ElevenLabs, OpenAI, Microsoft, Resemble AI
Emerged
2010s–2020s
Related
Text-to-Speech, Deepfake, AI Watermarking

Voice cloning is the use of artificial intelligence and machine learning to create a digital replica of a person's voice, allowing users to synthesize completely new speech in that voice from text.[2] A well-trained voice clone closely resembles the original speaker's accent, intonation, pitch, pace, and resonance.[2]

History

Voice cloning builds on decades of speech synthesis research, in which computers generate spoken language from text. Audio deepfake technology — also referred to as voice cloning — is an application of AI designed to generate speech that convincingly mimics specific individuals, often synthesizing phrases they have never spoken.[1] The platform 15.ai is credited as the first to popularize AI voice cloning in memes and content creation, and in January 2022 the first instance of speech-synthesis fraud occurred when a cryptocurrency company sold voice lines generated with 15.ai as NFTs without permission.[1] ElevenLabs, founded in 2022, launched its first voice model in January 2023, which could read any text in any voice, including clones of a user's own voice; demand grew rapidly for audiobooks, video dubbing, and content creation.[3] The company later introduced a "no-go" list of politician and celebrity voices after an ElevenLabs-made clone of Joe Biden's voice was used in a robocall during the 2024 U.S. Democratic primary, and it now requires consent verification for newly cloned voices and offers a free deepfake detector.[3]

Key Concepts

Voice cloning systems are typically built on neural text-to-speech models trained on recorded speech. Providers such as ElevenLabs distinguish between instant voice cloning, which replicates a voice from short audio samples, and professional voice cloning, which produces higher-fidelity replicas with consent verification.[2] Because even a short sample of a person's voice — from a social media clip, a voicemail, or a phone call — can be used to create a convincing clone, regulators have warned that scammers can clone a voice from just a few seconds of audio.[4] Mitigations include consent verification at cloning time, audio watermarking and deepfake detection tools, and verification practices such as agreed "safe phrases" for sensitive requests.[3][4]

Applications and Impact

Legitimate uses of voice cloning include audiobook production, multilingual video dubbing, accessibility tools, customer service voice agents, and education; ElevenLabs, for example, partnered with actor Eric Dane before his passing to recreate a digital copy of his voice for his family, and with the education platform PhysicsWallah to power AI tutoring.[2] The technology is also widely misused: the U.S. Federal Trade Commission warned in 2023 that scammers were using AI to clone voices in "family emergency" schemes, in which callers impersonate relatives to demand urgent money transfers.[4]

>See Also

References

🇲🇾Malaysian Context

Voice cloning scams have been reported across Malaysia. In May 2025, a woman in Selangor lost RM5,000 after falling victim to a voice cloning scam that used AI to mimic a relative's voice, and Malaysian media have documented deepfake-voice calls and WhatsApp messages in which scammers impersonate friends, family members, or colleagues to request urgent transfers.[5][6] In July 2026, a Sarawakian consultant reported that AI had copied his voice to trick his friends into sending money to scammers.[7] Authorities including Bank Negara Malaysia, the Royal Malaysia Police, and the Malaysian Communications and Multimedia Commission (MCMC) have repeatedly warned the public that banks and government agencies will never request personal information or payments by phone, message, or social media.[8] The use of cloned voices also raises issues under Malaysia's Personal Data Protection Act (PDPA), since voice data used to create clones may constitute personal data processed without consent.[9]

References

  1. [Speech synthesis — Wikipedia](https://en.wikipedia.org/wiki/Speech_synthesis)
  2. [What is voice cloning and how does AI voice cloning work? — ElevenLabs](https://elevenlabs.io/blog/what-is-voice-cloning)
  3. [How a tiny Polish startup became the multi-billion-dollar voice of AI — Forbes](https://www.forbes.com/sites/iainmartin/2025/12/01/how-a-tiny-polish-startup-became-the-multi-billion-dollar-voice-of-ai)
  4. [Scammers use AI to enhance their family emergency schemes — U.S. FTC](https://consumer.ftc.gov/consumer-alerts/2023/03/scammers-use-ai-enhance-their-family-emergency-schemes)
  5. [AI scams are getting real: cases happening in Malaysia — Malay Mail](https://www.malaymail.com/news/malaysia/2025/08/04/ai-scams-are-getting-real-here-are-the-cases-happening-in-malaysia-that-you-should-know-about/183459)
  6. ['It sounded just like my brother': how deepfake voices are fuelling money scams — Malay Mail](https://www.malaymail.com/news/malaysia/2025/08/04/it-sounded-just-like-my-brother-how-deepfake-voices-are-fuelling-money-scams/183345)
  7. [Sarawakian consultant claims AI copied his voice to trick friends into sending money to scammers — Malay Mail](https://www.malaymail.com/news/malaysia/2026/07/18/sarawakian-consultant-claims-ai-copied-his-voice-to-trick-friends-into-sending-money-to-scammers/228055)
  8. [Alert on SMS scam and fake website involving Bank Negara Malaysia name — BNM](https://www.bnm.gov.my/-/alert-on-sms-scam-and-fake-website-involving-bank-negara-malaysia-name)
  9. [PDPA AI compliance — AIWiki Malaysia](/wiki/pdpa-ai-compliance)