Cover Image for "I cloned my voice with artificial intelligence and even my wife doesn't notice the difference."
Mon Nov 18 2024

"I cloned my voice with artificial intelligence and even my wife doesn't notice the difference."

I think he/she likes it better this way.

Currently, the possibility of hearing your own voice pronouncing words you’ve never said is an unsettling experience. However, in this 2024 marked by advancements in AI, it’s not so surprising anymore. Technology has progressed to the point where it is possible to clone a person's voice and make it sound identical—a phenomenon that seems almost ordinary.

What is truly astonishing is how easy this procedure can be. There are numerous AI voice cloning options available online, many of which are free. These types of tools only require a brief training period of about 30 seconds. Once completed, it is possible to generate audio with the cloned voice without significant restrictions, raising concerns about a lack of security controls, which even allows for creating inappropriate or threatening messages.

While researching voice cloning options, it became clear that there are many alternatives available. Although many of these services require a monthly subscription, there are also options that offer free services. After trying out some of these alternatives, I found that, while they promised exceptional accuracy, some produced a robotic version that would fool no one. My goal was to create a clone of my voice that could confuse my wife.

Ultimately, I chose Speechify, which stood out for its ease of use and the quick 30-second training process. By creating a free account, I simply spoke into the microphone for the required time and was able to generate text that sounded in my own voice. Additionally, Speechify emphasizes a serious commitment to privacy, stating that it will never sell users' information.

When I tested the result with my wife, although she initially laughed upon recognizing it was coming from my computer, she admitted that the clone sounded realistic. On one hand, voice cloning presents advantages, such as eliminating common speaking errors. This can be particularly useful in podcast production, where a flawless presentation is desired. However, the potential for abuse is considerable, as it allows a fake voice to say almost anything.

On the other hand, the use of cloned voices also raises ethical and security dilemmas. Recently, the famous British presenter Michael Parkinson revealed that he would launch a new podcast using an AI version of his voice, which generated mixed reactions. While his estate supports the project, questions arise about using someone's voice without their consent. David Attenborough, a prominent figure in natural history programming, has expressed concern over the idea of his voice being cloned, finding it disturbing.

We live in an era where it is possible to generate podcast content without human intervention, posing significant challenges. The ability to clone voices, whether from deceased celebrities or ordinary individuals, leaves an uncertain future for public figures as well as for people in general.