How to set up and use Apple Personal Voice

Last checked against Apple's documentation on 19 July 2026, for iOS 26 and iPadOS 26.

Personal Voice is a feature Apple built into iPhone, iPad and Mac in 2023, aimed at people who are at risk of losing their ability to speak. You read some phrases aloud, your device trains a voice model on them, and afterwards apps can speak with a voice that sounds like yours instead of a stock synthetic one.

It costs nothing, it works entirely on your own device, and it takes one sitting. The hard part is knowing to do it early. A voice model is only as good as the speech you record, so the best day to bank your voice is the day you first wonder whether you should.

The one thing to take away: record it while speaking is still comfortable. Nobody regrets banking their voice too early. Many people regret waiting.

Before you start

A supported device. Apple's support article lists iPhone 12 or later, iPad Air (5th generation) or later, the recent iPad Pro models, and Macs with Apple silicon. The newest, fastest version of the recording flow runs on more recent hardware still.
iOS 17 or later, iPadOS 17 or later, or a recent macOS.
A passcode, Face ID or Touch ID set on the device. Without one, the option to create a Personal Voice does not appear at all. This trips up more people than anything else on this page.
A quiet room and about half an hour of unhurried time, even if the recording itself is shorter.
A supported language. See the next section, because this rules some people out.

Which languages Personal Voice supports

As of iOS 26, exactly three:

English (United States)
Mandarin Chinese (China mainland)
Spanish (Mexico), added in iOS 26

German is not supported. Neither is Spanish as spoken in Spain, French, Italian or any other language. If you speak one of those, you cannot bank a Personal Voice in your own language today, and we would rather tell you that plainly than let you find out forty phrases in. Apple has not announced further languages.

Rejoin speaks German and Spanish with high-quality built-in system voices, and you can pick and adjust one during setup. It is not your voice, and we will not pretend otherwise.

Recording your Personal Voice, step by step

1. Set a passcode first

If the device has no passcode, Face ID or Touch ID, the Personal Voice option is hidden. Set one in Settings → Face ID & Passcode (or Touch ID & Passcode) and come back.

2. Open Settings → Accessibility → Personal Voice

On a Mac it is System Settings → Accessibility → Personal Voice.

3. Tap Create a Personal Voice

Give the voice a name. If you record more than one over time, useful names help: "Dad, March" beats "Personal Voice 2".

4. Read the phrases aloud

Your device shows randomized phrases and you read each one. How many depends on your device. On the newest iPhones and iPads, Apple's faster flow uses ten phrases and takes under a minute. On older supported devices the original flow asks for 150 sentences, which Apple described as about fifteen minutes of recording.

Sit somewhere quiet. Apple's system screens out recordings that are too noisy, and the quality of the finished voice depends directly on the quality of what you record. You can use the built-in microphone, or an external microphone or headset if you have one. Speak the way you normally speak, at the pace you normally speak. You are not auditioning.

5. Pause whenever you need to

Tap Done, or simply close Settings, and your progress is saved. To carry on later, go back to Personal Voice, tap the voice you were creating, and tap Continue Recording. You can spread it across as many sittings as you like. Tiredness is a normal part of this, especially with ALS or MS, and stopping does not cost you anything.

6. Wait for the notification

Your recorded speech is processed securely on the device, and you get a notification when the voice is ready. On newer hardware this can be quick. When the feature launched, Apple processed the voice overnight while the device was charging, locked and on Wi-Fi, so if nothing has happened after a few minutes, plug the phone in, lock it, and leave it until morning.

7. Let your apps use it

Go to Settings → Accessibility → Personal Voice and turn on Allow Apps to Request to Use. On a Mac the wording is "Allow applications to use your Personal Voice".

Apps you approve can speak aloud with your voice, through the speaker or on a call. Apple is explicit that they cannot capture speech from Personal Voice: an app can use the voice, but it cannot take a copy of it away. The same toggle is where you revoke access.

If your speech has already changed

This is the part of Personal Voice that is most under-publicised, and for many of the people reading this it is the whole ballgame.

You do not have to read full sentences. Apple added an accommodation for people who have difficulty pronouncing or reading full sentences: you can create a Personal Voice using short, three-word phrases instead. If long sentences are already beyond you, this path may still work.

Two more things that help: you can pause and resume across as many days as you need, and you can create more than one voice on a device. If you are not sure whether today's recording is good enough, record one anyway. You can make another. What you cannot do is record a voice you no longer have.

If the result does not sound like you, that is disappointing and it is also not a verdict on you. Rejoin's built-in voices are warm and adjustable, and you can switch between them and your Personal Voice at any time.

Using your voice on your other devices

Turn on Share Across Devices and your Personal Voice becomes available on other devices signed in to the same iCloud account. This needs two-factor authentication on your Apple Account.

Doing this also stores a copy in iCloud, which is the only backup of your voice that exists. Apple states that Personal Voice is encrypted and stored on device so that only you can access it, and that the iCloud copy uses end-to-end encryption. Turn this on. Replacing a phone without it means recording everything again.

Things worth knowing before you commit

Deleting is permanent. Once you delete a Personal Voice it cannot be recovered.
Your Legacy Contact cannot access your Personal Voice. It does not pass on through Apple's Digital Legacy. If that matters to your family, it is worth talking about while you can.
Apple licenses it for personal, non-commercial use, made from your own voice, for a voice that sounds like you.
It is generated on your device, not on a server. Apple states that training and inference happen entirely on device.

Personal Voice and Live Speech

Once a Personal Voice exists, the fastest way to use it is Apple's own Live Speech: turn it on in Settings → Accessibility → Live Speech, pick your voice under Voices, then triple-click the side button, type, and tap Send. You can store a list of Favorite Phrases.

Live Speech is free, built in, and genuinely useful. It is a text field and a flat list of saved phrases, reached by a triple-click. That is the right tool for occasional use, and people manage entire conversations with it.

What it does not try to be is a communication app: there is no board of your phrases organised the way you think about them, no history, no routines, no layout designed for a tremor or a switch or an eye-gaze rig, and nothing that helps you keep up with the pace of a conversation. Apple positions third-party AAC apps as the answer to that, and explicitly allows them to use your Personal Voice.

Common questions

What languages does Apple Personal Voice support?

Three: English (United States), Mandarin Chinese (China mainland) and Spanish (Mexico). German is not supported, and neither is Spanish as spoken in Spain. Apple added Mandarin in iOS 18 and Spanish (Mexico) in iOS 26.

How long does it take to record a Personal Voice?

It depends on your device. On the newest iPhones and iPads, Apple's faster flow uses ten randomized phrases and takes under a minute. The original flow, still used on older supported devices, asks you to read 150 sentences and took Apple's stated fifteen minutes. Either way you can pause and resume as often as you like.

Can I create a Personal Voice if my speech has already changed?

Often yes. Apple added an accommodation for exactly this: if you have difficulty pronouncing or reading full sentences, you can create a Personal Voice using short, three-word phrases instead. You can also pause and resume across as many sittings as you need, and you can create more than one voice.

Which apps can use my Apple Personal Voice?

Apple's own Live Speech, and any third-party app you approve, once you turn on Allow Apps to Request to Use in Settings under Accessibility and Personal Voice. Approved apps can speak through your device with your voice, but they cannot capture speech from Personal Voice. Rejoin uses this mechanism, so your voice appears automatically once you allow it.

Is my Personal Voice stored on Apple's servers?

It is created and stored on your device, and Apple states that model training and inference happen entirely on device. If you turn on Share Across Devices, a copy is stored in iCloud with end-to-end encryption so it can be restored if you replace your device. That requires two-factor authentication on your Apple Account.

Where Rejoin comes in

Rejoin is an AAC app built around one problem: by the time you have typed an answer, the conversation has moved on. It listens alongside you, and when someone asks you something, three replies written in your own style are already waiting. You tap one, and it is spoken in your Personal Voice.

There is no export step and no file to upload. Once you have turned on Allow Apps to Request to Use, Rejoin asks for permission through Apple's own API and your voice appears in the list during setup. Setup takes about three minutes, and your first sentence is spoken aloud during it.

Everything you need in order to speak is free, forever: typing with your voice, your sayings board, routines, Display mode and history, all working offline. The AI layer that listens and drafts replies is the paid part, and it needs a connection. If a subscription ever lapses, Rejoin does not go silent.

Questions about banking a voice, from anyone, whether or not you use Rejoin: hello@rejoinvoice.com. We answer within a day.

Sources

Every factual claim on this page comes from Apple's own documentation: Create a Personal Voice, Use Live Speech, iOS feature availability, Apple Accessibility: Speech, and Apple Machine Learning Research. Apple changes these features. If something here has gone stale, please tell us.