Tom Cruise on TikTok, Trump having a baby with Kamala Harris, or even Barack Obama talking nonsense (Catherine Mackay, 2019)—celebrities are increasingly impacted by Deepfake technology.
Deepfake is a remarkable combination of AI voice generation, AI lip-syncing, AI image generation, and other advanced AI software. Together, they create a convincing replica, or “deep fake,” of anyone with enough digital content available online—which includes almost every celebrity or influencer.
In this blogpost, I will explain a new intriguing application that I discovered last year. Like many of you , I have seen these videos where we can see celebrities talking and saying obvious fake news. This intrigues me, and I did some research on the different technologies I could use for myself, and the voice generator was something very intriguing, to say the least.
Today, there are popular applications like ElevenLabs, which lets you create highly convincing voices or sound effects, and FakeYou, which mimics famous people or characters.
ElevenLabs has already been helpful for one of my academic projects. One challenge when recording a video is that without professional equipment, the audio quality can seem amateur. But with this AI software, you can generate a professional-sounding voice and easily integrate it into your workflow. The software allows for a range of adjustments, such as making the voice more enthusiastic or angry, or changing the speed of the narration (as you can hear in Audio Clip 1). The platform even provides background sound effects—like in the Audio Clip 2 where I requested a bustling cityscape with people chatting and cars passing by.
On the other hand, FakeYou is more about fun at the moment. The voices aren’t as clear and don’t offer as much customization as ElevenLabs, but it lets you choose a character or personality and have them say whatever you want. In Audio Clip 3, for example, I used the voice of SpongeBob.
In conclusion, I encourage you to explore these tools—whether it’s just for fun with FakeYou or for more professional uses with ElevenLabs. Both offer exciting possibilities, and you can create a professional voiceover without breaking the bank!
Reference
Catherine Mackay. (2019, September 16). DeepFake Obama JordanPeele Short [Video]. YouTube. https://www.youtube.com/watch?v=on7wVajA9Qs
Ahahah, I actually laughed getting to the SpangeBob part… but nevertheless the Obama clip is mind-blowing. I was familiar with deepfake all over the place, affecting even elections, but I’ve seen it with my own eyes for the first time. (or did I not? :o) In any case, I’m off to try those websites and I can’t stop thinking about potential applications now.. For example, AI reading audiobooks, singing pop songs, reading podcast scripts, generating public transportation or other public messages… Oh, man. Personal assistants surely.. I can see that soon we’ll live in a world half-populated by AI humanoid companions.
Thank you for your feedback! I’m glad my audio clip made you laugh! There are indeed many exciting applications for this technology, and it’s an incredible time to witness what we’ll achieve with it. The idea of a “half-populated” world is also fascinating. Some conspiracy theorists even suggest the “dead internet” theory, where much of what we see online is generated by bots, not real humans. While it may seem far-fetched now, this concept could become a reality in the near future.
Hello Samuel, I just read your blog and I’m very intrigued by the technologies you present. I was wondering what you think is the biggest threat coming from technologies like these. Is identity theft the biggest crime you could commit with these means, or are you more worried about the potential for fake news in our already media-sceptic policial environment? Let me know what your thoughts are.
Thank you for your feedback. You’ve raised excellent points, and I agree that they all pose significant potential threats. It’s hard to say which one will prove to be the most harmful. As with any new technology, it’s crucial to establish safeguards, and only time will tell if our institutions and the companies developing these technologies will be capable of protecting us from future challenges.
Your article is very intriguing to me. I am very fascinated by your optimism regarding deepfakes. They scare me a lot, especially how much influence they could potentially have on the internet and people’s opinions regarding certain topics or people. I am, therefore, very pessimistic about the idea of using and experimenting with deepfakes in the future, as I am convinced they could provide more harm than good. My personal opinion is that there should be a clear set of rules for irresponsible use of deepfakes. However, I am curious what your response to this concern would be.
Thank you for your feedback! Your concerns are completely valid—deepfakes are indeed a concerning technology. While we can currently recognize them with some awareness, advancements in the field will make it increasingly difficult to distinguish between deepfakes and reality. Some companies, like TikTok and Meta, already require AI-generated content to be flagged, and the EU is actively working on regulations to address this issue. Unfortunately, while deepfake technology stems from fascinating innovations, it has the potential to be misused. That said, the underlying technology holds great promise and could shape our future in positive ways if used responsibly.