How to Clone Your Own Voice on an iPhone

Apple’s Personal Voice accessibility feature is built with privacy safeguards, but AI-driven voice mimicry has a creepy side

Edward C. Baig,

AARP

Comments

En español

Published September 11, 2023

an illustration of a smartphone on a purple field emanating a sound wave with a microphone on the screen

GETTY IMAGES

A person dear to you has been diagnosed with amyotrophic lateral sclerosis (ALS) or another devastating disease that puts their ability to speak at risk.

If only you could bank their voice, you think. It’s now possible.

The Personal Voice accessibility tool that arrives this month with Apple’s free iOS 17 software upgrade for iPhones allows people to create an authentic-sounding, albeit slightly robotic replica of their voice — before they lose the ability to speak. The tool, which healthy people can take advantage of as well, works in tandem with another new iOS 17 feature called Live Speech.

The feature lets people type phrases they want spoken aloud, which can be voiced with the synthetic Personal Voice. Apple built the feature with privacy and security protections.

For now, Google doesn’t have an equivalent feature for Android.

“At the end of the day, the most important thing is being able to communicate with friends and family,” ALS advocate Philip Green of the nonprofit Team Gleason organization said in a statement released by Apple. “If you can tell them you love them in a voice that sounds like you, it makes all the difference in the world.”

Apple won’t reveal its plans, but similar technology could be leveraged someday to preserve grandparents’ or other relatives’ voices, left behind for a younger generation after they’re gone.

The idea of voice cloning is unsettling to some

For all of its promise as a technology, voice cloning is understandably worrisome, especially as leaps in generative artificial intelligence produce convincing fake voices. Readily available software tools make it relatively easy to mimic another person’s voice in minutes, often through capturing audio recordings from podcasts, social media and the web.

The creepy part: Scammers are already exploiting such tools. You may have read about incidents where malicious impostors pose behind a loved one’s cloned voice to dupe victims into surrendering personally sensitive information or cash.

The criminals can use as little as three or four seconds of a clip to replicate your voice, says James E. Lee, chief operating officer of the San Diego-based nonprofit Identity Theft Resource Center.

“Identity criminals don’t need it to be perfect,” he says. “They just need something that’s close enough because … the most common thing we see is some sort of relationship scam.”

Lee points to another potentially significant risk: As organizations turn to voice to authenticate an account holder, someone with a good enough clone of a person’s voice could access their account.

Politicians have reasons to be concerned as well. Cloned voices can be used to spread misinformation and deceive voters, similar to the way that deepfake pictures and videos can mislead the public. And AI will make fakes even harder to detect.

Faux voices are being employed in other ways that, at the very least, spur debate about ethics and the proper use of the technology. It conjures up memories of the 1970s’ “Is it live or is it Memorex?” commercials.

In April, an AI-generated song called Heart on My Sleeve used cloned vocals from pop musicians Drake and The Weeknd. The track went viral before Spotify and other streaming platforms removed it when music label Universal Music Group cried foul over copyright violations.

The song, created by someone called Ghostwriter, was submitted for Grammy consideration, according to The New York Times. The Recording Academy subsequently declared it ineligible. Ghostwriter recently surfaced again with an AI-generated track featuring voices that sound like rappers Travis Scott and 21 Savage.

Apple says it keeps your Personal Voice private

Apple has several measures to ensure that someone else cannot access your Personal Voice. For starters, your Personal Voice is encrypted and stored securely on your device.

You must unlock the device with FaceID, TouchID or a pass code to use it. An Apple device without a pass code can’t create a Personal Voice.

You generate the Personal Voice on your own device, not through importing audio clips off podcasts, social media or elsewhere. Instead, you must read 150 randomized phrases aloud, a deliberate drill that takes a while.

The finished voice remains on your own phone or tablet, though a user can sync the voice with an iPad or Mac, provided each device is running the upcoming iPadOS 17 and/or macOS Sonoma software, and they’ve opted in via iCloud. Apple is also opening Personal Voice to developers, but these companies have to ask your permission to use it in their apps on your phone.

“The only way that that can be misused by somebody who has ill intent is if they have control of your device,” Lee says. “It is a risk, but a pretty low risk.”

How to create your Personal Voice

1. Make sure your iPhone or iPad is running iOS 17 or later or iPadOS 17 or later.

2. Go to Settings | Accessibility and scroll down to Personal Voice in the Speech section.

3. Tap Create a Personal Voice | Continue.

4. As you prepare to record, find a quiet room with minimal echo and background noise.

5. Over the next 15 minutes or so, you’ll be asked to read 150 randomized phrases aloud. Among them:

“Her style of painting shows the influence of the French artist.”
“She played an important part in fiscal and foreign policy.”
“Have you ever sneezed and coughed at the same time?”

Speak naturally and hold the phone 6 to 12 inches away. After you read a phrase, a new one appears, but you can save and close your session any time and resume later.

6. Tap Continue once you’ve successfully recorded all the phrases. Your Personal Voice will be created securely on your phone, a process that can take hours, according to Apple

That was the case when I recorded my Personal Voice as iOS 17 was in its beta testing phase. My voice sounded a lot like me, but the cadence was sometimes off.

Leave your phone charged and locked, which is the only condition under which the voice will be generated, Apple says. You can track the progress in Settings and will receive a notification when the process has been completed. You also can assign it a name.

How do you use your Personal Voice?

1. Once you created your Personal Voice, open Settings again | Accessibility | Live Speech.

2. Tap the Live Speech switch to make it green, rather than gray.

3. Tap English (US) under Voices.

4. The Personal Voice you created should appear on top of the other voice options listed, which are numerous synthesized voices from Apple. Tap your Personal Voice, which will carry whatever name you gave it. A check mark indicates it is the voice you’ve selected.

5. Triple-click the side button on the iPhone to summon the Live Speech prompt.

6. Using the iPhone keyboard, type the words you want spoken aloud in the space provided and tap Send to hear those words in your voice.

%{postComment}%

Edward C. Baig covers technology and other consumer topics. He previously worked for USA Today, BusinessWeek, U.S. News & World Report and Fortune, and is author of Macs for Dummies and coauthor of iPhone for Dummies and iPad for Dummies. Follow him on LinkedIn; Threads; and X, formerly known as Twitter.