AI Voice Cloning and Voiceovers: Clone Your Own Voice Safely (and Spot the Scams)

Short answer: For most videos, a stock AI voice is enough, and many tools offer them free. If you want the voiceover to sound like you, record one to two minutes of clean speech and upload it to a voice-cloning tool such as ElevenLabs, whose cheapest paid plan includes Instant Voice Cloning. Only clone your own voice, or someone's voice with their written permission.

This guide is for creators who want narration for videos, courses or a podcast, and for anyone worried about scammers faking a relative's voice.

Do you need a voice clone or a stock AI voice?

A stock voice is a ready-made synthetic voice: you type a script and pick a voice. A clone is a voice model trained on recordings of one real person, so the output sounds like them.

In our view, start with stock voices. Clone your voice only when your audience already knows how you sound, for example on a channel or course you host. You can compare more tools in our speech API directory.

Never clone a voice without the speaker's permission. ElevenLabs asks you to "confirm that you have the right and consent to clone the voice" before it saves a clone. Faking someone's voice can also break platform rules and, depending on where you live, the law.

How do you clone your own voice with ElevenLabs?

As of November 2026, ElevenLabs' pricing page lists Instant Voice Cloning on the Starter plan ($6 a month, with $1 for the first month) and above. The Free plan does not list cloning. The steps below follow ElevenLabs' documentation.

1. Sign in to ElevenLabs on a paid plan.
2. Select Voices in the left sidebar.
3. Click the plus icon.
4. Choose Instant Voice Clone.
5. Upload or record your audio, following the on-screen prompts.
6. Name and label your voice.
7. Confirm you have the right and consent to clone the voice.
8. Click Save voice.
9. To use it, open Voices, then My Voices, and click Use voice.

This is ElevenLabs' own tutorial, so it naturally features their product. It walks through recording a sample and generating speech with the clone.

Your voice recordings are personal data. Read ElevenLabs' privacy policy before uploading, and delete clones you no longer use.

How do you record a clean voice sample?

ElevenLabs says the clone copies everything it hears, including background noise and mouth clicks. So the recording matters more than the tool.

1. Record in a small, soft room: curtains, cushions or a wardrobe full of clothes.
2. Turn off fans, air conditioning and notifications.
3. Keep the same distance from the microphone throughout.
4. Speak in the tone you want the clone to use, at an even pace.
5. Record about 1–2 minutes in total; ElevenLabs advises against going past 3 minutes.
6. Export as MP3 at 192 kbps or higher, as ElevenLabs' FAQ recommends.
7. Listen back on headphones and re-record anything with echo, clicks or noise.

Consistency helps more than drama. ElevenLabs notes that audio with big swings in pitch and volume gives less predictable results. If you are recording for a show, our guide to making a podcast with AI covers the rest of the setup.

Can you make an AI voiceover for free?

Yes, but "free" usually comes with limits on commercial use. Check the licence before you publish anything that earns money.

• ElevenLabs Free: 10,000 credits a month as of November 2026, and text-to-speech costs about one credit per character. ElevenLabs states the Free plan "cannot be used for any commercial purpose". If you publish free-plan audio, you must credit ElevenLabs by putting "elevenlabs.io" or "11.ai" in the title.
• Piper: A fast, local text-to-speech engine that runs on your own computer, so your script never leaves it. The current code (the OHF-Voice/piper1-gpl repository) is GPL-3.0. Each voice has its own dataset licence in its model card. For example, the popular US English "lessac" voice was trained on a dataset licensed for research only, with commercial use excluded.
• Open-source cloning models: Check the weights' licence, not just the code. Coqui's XTTS-v2, for example, is under the Coqui Public Model License, which "allows only non-commercial use of a machine learning model and its outputs."
• CapCut: CapCut's own guide shows text-to-speech in its mobile app. You add your script with Text, tap Text to speech, pick a voice, then adjust speed and pitch. The guide doesn't say whether every voice is free, so check inside the app.

Canva also has AI voice features. Check the current plan limits in Canva's help center before you rely on them.

Did you know? The model card for Coqui's open-source XTTS-v2 advertises "voice cloning with just a 6-second audio clip." That's one reason the FTC warns that a short clip from a social-media post can be enough for a scammer.

How do you protect your family from voice-clone scams?

The typical scam is a panicked call from a "relative" who needs money now. The US FTC warns: "Don't trust the voice." These steps come from FTC and FBI advice.

1. Agree on a secret family safe word that a caller must say.
2. If a call asks for money, hang up.
3. Call the person back on a number you already know.
4. If you can't reach them, call another family member or a friend.
5. Refuse requests for wire transfers, cryptocurrency or gift-card codes.
6. Make social accounts private and limit public clips of your voice.
7. Report it: in India, call 1930 or file at cybercrime.gov.in; in the US, use ReportFraud.ftc.gov.

If you've already sent money, report it at once. India's Ministry of Home Affairs says its 1930 fraud-reporting system is built for "quick reporting of financial cyber frauds" and aims to "prevent flow of money siphoned off" from victims.

Fake video is the same problem with pictures. Our guide on how to detect deepfakes and AI images covers fake photos, videos and video calls.

Frequently Asked Questions

How does AI voice cloning work?

A model learns the sound of one person's voice from sample recordings, then speaks any new text you type in that voice. Better, cleaner samples give a more convincing clone.

Can I clone my voice for free?

ElevenLabs' Free plan doesn't list voice cloning as of November 2026; cloning starts on the Starter plan. Open-source cloning models are free to download, but many, such as XTTS-v2, allow non-commercial use only.

Is AI voice cloning legal?

Cloning your own voice is generally fine. Cloning someone else needs their clear permission, and tools like ElevenLabs make you confirm consent. Laws differ by country, so get legal advice for commercial projects.

Can I make an AI voiceover in CapCut or Canva?

CapCut's guide documents a Text to speech option for text in its mobile app. Canva has AI voice features too; check its help center for current plan limits.

How can I tell if a phone call is an AI voice clone?

You often can't tell by ear, so don't try. Hang up, call the person back on a number you know, and use your family safe word.

Sources

• ElevenLabs pricing
• ElevenLabs docs: Instant Voice Cloning
• ElevenLabs help: Can I publish the content I generate on the platform?
• Piper (OHF-Voice/piper1-gpl) on GitHub
• Piper voices: lessac model card and Blizzard 2013 Lessac dataset licence
• Coqui XTTS-v2 model card and licence
• CapCut: How to add voice over in CapCut
• FTC: Scammers use AI to enhance their family emergency schemes
• FBI IC3: Criminals use generative AI to facilitate financial fraud
• I4C (Ministry of Home Affairs): National Cyber Crime Reporting Portal and 1930 helpline

Reviewed on November 3, 2026. Plans, credit amounts, prices and app menu labels change often; check the vendors' pages before you buy or publish.