Subscribe to our newsletter:

Samsung Bixby Breakthrough: AI Will Clone Your Voice

16 Mar 2023 News
rss subscribe
RSS Subscribe
Apps on sale

Developing technology is changing all industries, and the fantasy worlds we see in science fiction movies are becoming reality day by day. You can see the effects of this even on a site like — you can now play against real dealers and attend dozens of different live game shows from the comfort of your home. Digital assistants are among the best examples of what modern technology can do because they now work with artificial intelligence and offer incredible things. For example, Samsung Bixby can create an AI-generated copy of your voice. Wondering how is this possible and how can it be used?

Cloning Your Voice Using AI

Bixby is the name of the digital assistant Samsung has been using since 2017. It was developed as an alternative to Apple's Siri, and you can find it on almost all Samsung mobile devices. It works exactly as you'd expect. You can give Bixby voice commands and have it launch an app or make a call. Also, it’s possible to create appointments with it, add reminders, and even have it buy plane tickets for you. These are impressive features, but not unexpected, as all digital assistants currently on the market can do them.

However, Samsung is trying to make a difference by getting support from artificial intelligence. The "Bixby Text Call" service, recently launched in Korea, creates an AI-generated copy of your voice so you can answer calls “personally.” The goal is to tell your callers that you're busy with your own voice when you can't pick up the phone. But you don't have to record a message to do that because artificial intelligence does it for you! Callers hear from your voice why you can't pick up the phone at that moment.

To take advantage of this service, you must use Galaxy S23, S23+, or S23 Ultra and update to One UI 5. The “Text Call” feature currently only supports Korean but will start supporting English in March 2023. This feature works roughly like this:

  • You start Bixby and record your voice. It doesn't matter what you say — artificial intelligence just wants to hear the words and sentences (their tones) with your voice.
  • Once the recording is complete, Bixby analyzes it and creates a copy of your voice. Now it can use this copy to create different voice messages explaining why you can't pick up the phone. The caller will not understand that the voice is not yours.

The Technology Already Has Its Rivals

Samsung also announced that it plans to use the service for more than just phone calls. No one knows what exactly this will entail, but it is safe to say that Bixby already has competitors that can do the same:

  • Microsoft uses VALL-E to create sound copies and claims it can better reflect emotional tones.
  • Amazon wants to use this technology not to answer calls, but to make you feel like your loved ones are still alive. You can talk to a deceased relative and hear their voice again.
  • With Duplex AI, Google aims for artificial intelligence to complete simple tasks using your voice. For instance, your digital assistant can call your favorite restaurant and make a reservation with your voice.

These impressive features and projects seemed like a dream until recently. However, they all have some "creepy" aspects too. It may not be very healthy to talk to a deceased relative and hear their voice again. Also, AI can do a lot of things with your audio copy without asking. What if it gets hacked to call the bank to transfer all the money from your account to another one? This is a possible scenario! And your digital assistant already has all your passwords/biometric data to “get to know you better.” If it can also use your voice, there's no limit to what it can do.

But even assuming it's a "hack-proof" technology, there's almost no way to counter the security risks posed by third parties. Anyone who can access a recording of your voice (or record your voice) will be able to use it in a highly convincing deep fake video and have you say anything through artificial intelligence. This has already happened! Not long ago, trolls gathered on the 4Chan site and posted several hate-speech videos using public audio recordings and free services. In these videos, you could see many people, from cartoon characters to famous political figures, saying pretty offensive things. These were deep fake videos, but they looked much more believable thanks to the audio cloning technology.

Despite security concerns, it looks like voice cloning will soon become a part of all digital assistants. What you can do with it is almost limitless, and there isn't much we can do except hope that security systems will evolve as quickly as this technology.

Share this article: