ChatGPT and your voice

Can ChatGPT clone my voice?No, but it can write what your clone says

ChatGPT Voice speaks in preset voices only; OpenAI’s help page offers no way to clone your own. OpenAI does make custom voices from a recording in its API, but only for eligible customers through sales. To hear your own voice read text, clone it in a voice tool such as ElevenLabs or oVideo Voice Clone, and use ChatGPT to write or translate the script.

ChatGPT: preset voicesGemini, Grok, Canva answeredChecked November 20, 2026

Updated

Short answer

Can ChatGPT clone my voice?

No. In the ChatGPT app you choose one of OpenAI’s preset voices, such as Arbor, Juniper or Sol, and there is no option to clone your own. OpenAI’s developer platform can create a custom voice from a consent recording and a 10 to 60 second sample, but its guide says that is limited to eligible customers who contact sales. The practical route is a voice cloning tool: oVideo Voice Clone takes a recording or a video of you, is ready in about a minute, and reads anything you type in 90+ languages as an MP3.

  • ChatGPT app: preset voices only
  • OpenAI API: custom voices for eligible customers
  • Voice Engine (2024): previewed, not widely released
  • Your own voice today: a voice cloning tool

OpenAI’s help center and developer docs read on November 20, 2026.

What OpenAI offers

ChatGPT and voices, question by question

Which voices does ChatGPT have?

Arbor, Breeze, Cove, Ember, Juniper, Maple, Sol, Spruce and Vale, plus extra voices for Live in Brazilian Portuguese, French, German, Hindi, Japanese, Korean, Mexican Spanish and English, by OpenAI’s help page. You can ask it to speak faster, slower or in another language.

Is a custom voice coming to ChatGPT?

The help page says nothing about it. It does say Advanced and Standard voice retire on December 11, 2026, when everyone moves to Live.

What can developers do?

OpenAI’s API makes voices two ways: from a text description (open to all API customers, for GPT-Live), or from a consent recording plus a 10 to 60 second sample of the same person, limited to eligible customers through sales, at most 20 per organization. Children’s voices are not allowed.

What was Voice Engine?

A model OpenAI previewed on March 29, 2024 that made speech resembling a speaker from one 15 second sample. OpenAI said it was choosing to preview but not widely release it.

Read on help.openai.com, developers.openai.com and openai.com on November 20, 2026.

Can Gemini, Grok or Canva clone my voice?

The other assistants, from their own pages

Can Gemini clone my voice?

Yes, inside an avatar. Google’s help page says you record your face and voice on your phone to make an avatar in the Gemini app, then use it in images and videos that look and sound like you; making a video with it needs a Google AI plan. Limits: 18 or over, English only, not available in the EEA, Switzerland or the UK. Developers can replicate a voice with the Gemini API on a paid tier, from a 10 to 30 second clip plus a spoken consent statement, and every output carries a SynthID watermark.

Can Grok clone my voice?

Yes, for developers. xAI’s Custom Voices clone your voice in its developer console from about one to two minutes of speech; you read a passphrase so it can check the voice is yours, and its news post says you cannot clone someone else’s voice. It offers 30 free clones, 25+ languages and speech at $15 per million characters. Cloning inside the Grok chat app: not stated.

Can Canva clone my voice?

Yes, through the AI Voice Cloning app in Canva’s Apps Marketplace. Upload a WAV or MP3 of 10 seconds to 5 minutes, up to 15 MB, or record one, write the script and add the audio to your design. Canva’s page says over 130 languages and asks you to always disclose that the voice is AI generated; the app page says it is free with any Canva plan.

Can ElevenLabs clone my voice?

Yes. Its pricing page lists Instant Voice Cloning from the Starter plan at $6 a month and Professional Voice Cloning from Creator, while its voice cloning page says you can clone your voice for free in seconds. It is the engine behind oVideo Voice Clone.

Read on support.google.com, aistudio.google.com, x.ai, canva.com and elevenlabs.io on November 20, 2026.

How to do it instead

Clone your voice and let ChatGPT write the words

The steps on oVideo Voice Clone. On ElevenLabs the flow is similar: upload a sample, confirm consent, type, download.

  1. 01

    Record or upload

    Record in the browser, or upload an audio file or a video of you talking (MP4, MOV and others, up to 100 MB). At least 10 seconds; 30 seconds to 2 minutes is best.

  2. 02

    Confirm it is your voice

    Tick “This is my voice, or I have permission to use it.” Background noise is reduced and the voice is ready in about a minute.

  3. 03

    Ask ChatGPT for the script

    Use a prompt below to write, shorten or translate the text. Write it in the language you want to hear.

  4. 04

    Paste and generate

    Paste up to 5,000 characters. The credits show before you click: 25 per started 30 seconds of speech.

  5. 05

    Download the MP3

    Play it or download it; your last 20 are listed. To put it on a video, add the MP3 in any editor.

ChatGPT prompt: Write it in your voice

“Write a 45 second intro for my cooking channel in my own words. I talk casually, short sentences, a little dry humour. No hashtags, no emojis, nothing I would not say out loud.”

ChatGPT prompt: Translate for your clone

“Translate this script into natural Mexican Spanish for a voiceover. Keep it spoken, not formal, and keep names as they are. Give me only the Spanish text.”

ChatGPT prompt: Fit a time limit

“Shorten this to about 30 seconds when read aloud at a calm pace, roughly 75 words. Keep the first and last sentence.”

ChatGPT prompt: Make it easy to say

“Rewrite this so it sounds natural when spoken: spell out numbers, add commas where I would breathe, avoid tongue twisters.”

Compared, November 20, 2026

Which assistants and tools clone your voice

What each company’s own pages say, read on November 20, 2026. oVideo is our product, so read its column with that in mind.

FeatureoVideo Voice Clone (ours)ChatGPT appOpenAI APIGemini appGrok (xAI API)Canva
Clones your own voice?YesNo; preset voicesYes, for eligible customers through salesYes, as part of your avatar (face and voice)Yes, in the developer console and APIYes, with the AI Voice Cloning app
What you giveA recording, or an audio or video file; at least 10 s, 30 s to 2 min bestNot offeredA consent recording and a 10 to 60 s sampleA face and voice recording on your phoneAbout 1 to 2 min, plus a spoken passphraseWAV or MP3, 10 s to 5 min, up to 15 MB
What you getAn MP3 of any text you typeSpoken chat repliesSpeech in your appsImages and videos with your avatarSpeech over the APIAudio added to your Canva design
Languages90+ (Eleven v4, per ElevenLabs)Not applicableNot stated on the guideAvatars in English only25+Over 130, by its page
Where it worksBrowser, phone or computerChatGPT app and webYour own codeNot in the EEA, UK or SwitzerlandYour own codeCanva editor
PriceFrom $10/mo; a voice is 50 credits, speech 25 per started 30 sNot applicableThrough salesA Google AI plan to make avatar videos30 free clones; speech $15 per 1M charactersFree with any Canva plan, by the app page

Sources: help.openai.com ChatGPT Voice, developers.openai.com custom voices, support.google.com Gemini avatar and Google Account avatar pages, aistudio.google.com voice replication, x.ai Custom Voices and voice cloning, canva.com AI voice cloning and its app page.

Which one should you pick?

Pick by what you want to hear

  1. 1A chat assistant that talks back: ChatGPT Voice, in one of its preset voices.
  2. 2A video of you that looks and sounds like you, in English, outside Europe: a Gemini or Google Flow avatar.
  3. 3Your voice inside an app you are building: ElevenLabs, xAI Custom Voices or the Gemini API.
  4. 4Your voice on a Canva design: Canva’s AI Voice Cloning app.
  5. 5MP3s in your own voice, in any of 90+ languages, from a recording or a video of you: oVideo Voice Clone, with ChatGPT writing the script.

Checklist

What to check before you clone your voice, answered for oVideo

Is it only my voice?

You tick “This is my voice, or I have permission to use it.” ElevenLabs, which runs the clones, blocks famous voices.

How much audio?

At least 10 seconds of speech; 30 seconds to 2 minutes is best; the first 3 minutes are used.

Which languages?

90+, the languages ElevenLabs Eleven v4 speaks. Per ElevenLabs, your accent stays in your own language and other languages come out fluent rather than with your accent; we have not measured it.

Is it free?

No. Free accounts can record; creating a voice needs Starter ($10 a month, 400 credits) or higher.

What if it fails?

Credits are refunded automatically.

Can I delete it?

Yes. Deleting the voice removes it and your sample. MP3s you made are separate, so keep your downloads.

oVideo facts checked in the app on November 20, 2026.

Honest limits

What oVideo Voice Clone does not do

It does not talk back like ChatGPT: it turns text you type into an MP3. It does not sing, has no emotion or speed sliders, no voice changer and no dubbing with lip sync.

Its cloned voice works in the Voice Clone studio only; oVideo’s Add AI Voice, Story to Video and Shorts use their own voice lists for now. To put your voice on a video, download the MP3 and add it in any editor.

Clone only your own voice or one you have permission for.

Commercial use of the audio you make on oVideo is allowed under the Terms, provided you had the right to use the voice you cloned.

Answers

Frequently asked questions

Can ChatGPT clone my voice?
No. ChatGPT Voice lets you pick from preset voices such as Arbor, Breeze, Cove, Ember, Juniper, Maple, Sol, Spruce and Vale; its help page (read November 20, 2026) offers no custom or cloned voice. OpenAI’s API can make custom voices from a recording, but only for eligible customers through sales.
Can ChatGPT replicate my voice?
Not in the ChatGPT app. OpenAI previewed a model called Voice Engine in March 2024 that needed a 15 second sample, and said it was choosing to preview but not widely release it. To replicate your voice today, use a voice cloning app and let ChatGPT write the script.
What AI can clone your voice?
ElevenLabs, oVideo Voice Clone, HeyGen, Fliki, CapCut, Canva’s AI Voice Cloning app, Cartesia, Fish Audio and Speechma all clone your own voice, by their own pages. Google’s Gemini avatar and xAI’s developer console also do, with limits. See our best voice cloning apps comparison.
Can Gemini clone my voice?
Partly. The Gemini app makes an avatar from a recording of your face and voice, for images and videos that look and sound like you; it needs you to be 18 or over, works in English only and is not available in the EEA, Switzerland or the UK. Developers can also replicate a voice with the Gemini API on a paid tier, with a spoken consent statement.
Can Grok clone my voice?
Yes, for developers. xAI’s Custom Voices clone your voice in its developer console and API from about one to two minutes of speech, after you read a passphrase, with 30 free clones and 25+ languages. Its pages do not mention cloning inside the Grok chat app.
Can Canva clone my voice?
Yes, with the AI Voice Cloning app from Canva’s Apps Marketplace: upload a WAV or MP3 of 10 seconds to 5 minutes, up to 15 MB, type a script and add the audio to your design. The app page says it is free with any Canva plan, and Canva asks you to disclose that the voice is AI generated.
Is it illegal to clone my own voice?
Cloning your own voice is the allowed case on every tool we read. Cloning someone else without permission breaks the rules of OpenAI, Google, xAI, ElevenLabs and oVideo. Voice Clone also needs the speaker’s own permission, so it is not for recreating someone who has died.
Why did ChatGPT talk back in my voice?
People post about this on Reddit and OpenAI’s forum, but OpenAI’s help page does not describe any feature that clones your voice in ChatGPT. It lists preset voices only. We cannot confirm individual reports.
Does ChatGPT keep my voice recordings?
OpenAI’s help page says audio clips from Live and Advanced Voice are stored with the chat and retained for 30 days, deleted within 30 days when you delete the chat, and used for training only if you choose to share them.
How do I make ChatGPT read text in my voice?
Ask ChatGPT to write or translate the text, then paste it into a voice cloning app that has your voice. On oVideo Voice Clone that is up to 5,000 characters in any of 90+ languages, returned as an MP3.
Is there a free way to clone my voice?
Speechma is free with no account; Canva’s cloning app says it is free with any Canva plan; Grok’s console offers 30 free clones for developers. oVideo has no free voice clone; plans start at $10 a month.
Is my voice used to train AI models?
oVideo has opted out of ElevenLabs model training (since November 20, 2026), so voices you clone are not used to train ElevenLabs models. Your sample is stored privately and deleted together with the voice when you delete it.

Related

Sources, checked November 20, 2026

Hear your own voice read your script

Record or upload a clip of you, tick that it is your voice, paste the text ChatGPT wrote. An MP3 in about a minute, in any of 90+ languages.