Lights

Comments

The curious case of the AI Carlton Kirby

When your voice puts food on the table, and someone else is using it for free, what happens next?

Jonny Long
by Jonny Long 18.12.2024 Photography by
Carlton Kirby/X
More from Jonny + EscapeCollective Paywall Badge

If you’ve ever wondered whether Carlton Kirby is the same off-air as he is on his Eurosport commentary of bike races like the Tour de France, I can confirm after 10 minutes on the phone with him that yes, the experience certainly lives up to the bill.

He apologises for his croakiness, coming off the back of a 20-leg tour that occupies him during the initial months of the road off-season before he hunkers down for the deepest depths of winter. But first, he’s agreed to give his thoughts on a YouTube video I’d emailed to him over a week ago.

Clicking that link, Kirby was presented with his own voice blaring out over his laptop’s speakers. But this wasn’t a clip from any official Eurosport account, or even some race footage that had been ripped and re-uploaded.

Instead, it was his voice soundtracking videos titled “Tour de France 2025 is BUILT To DESTROY Tadej Pogačar” and “Remco Evenepoel’s CRASH Just RUINED the 2025 Tour de France”, which both live on a YouTube channel simply called Watts

The videos, usually around eight minutes in length, are fast-paced and well-edited, a series of photos interspersed with graphics to keep the viewer’s attention, which is a must if you want the YouTube algorithm to continue to serve up your videos to a proportion of the 100 million people who use the platform every day.

This style of video (photo slideshow plus snappy graphics and voiceover) is widespread on the platform, forming part of a concept commonly known as a “faceless YouTube channel,” where the person making the videos doesn’t have to show their face or voice if they don’t want to. Instead, they work as director, producer, animator, writer, and editor. 

Created in August 2023, the Watts channel initially used footage that was subsequently claimed for copyright infringement, and so now a total of 13 videos exist that have racked up 323,000 views, all featuring a version of Carlton Kirby’s voice.

Presently, Kirby doesn’t seem too concerned about what’s going on. This may partly be because there used to be an American website that had also ripped off and used his voice, and so this is nothing new. It may also be that empty question: what one can actually do about this?

Google “how to make an AI voice of someone” and you will find various websites offering their services. “How to Make Your Own AI Voice Model [Easy]” says one, “Free AI Voice Cloning In 30 Seconds! No Sign-up Required” reads another. The first instance of voice cloning technology dates back to 1998 but the outset of the 2020s saw various programmes become more widely accessible and used.

“A team of people aren’t required to create a voice,” Kirby points out. “It’s just a guy from a laptop.” Meaning anyone else could pop up tomorrow and do the same thing as Watts is doing.

“I don’t know where the line lies,” he continues, “as a humble busker in the world of entertainment, someone’s come and nicked my change.”

But that line proves, at least for now, that an infinite number of AI generators probably couldn’t whip up a 100% realistic Carlton Kirby replacement.

“My work is between brain and larynx,” he explains. “All I hope is my stream of consciousness can’t be mimicked. Although, if anyone mentions sausage factories, I’ll need to reconsider …”

Eurosport, his main employer, hasn’t said anything to him about it yet, but he does have some concerns outside of his work. For instance, the worry that someone could try to call up his bank using an AI-generated version of his voice and gain access to his accounts.

“At the moment I’m kind of treating it as an homage,” he joked before adding, “having your voice out there is one thing but having your voice out there saying things you’ve never said is another, not least because it could have legal ramifications for me. [For instance] if someone ripped off [Tour de France organiser] ASO footage and put my voice on it …” 

Another similar channel called “Cycling Highlights” has videos often centred on the topic of doping, with titles such as “Cycling World Champion Was Caught Doped” and an accompanying thumbnail image of Tadej Pogačar. Another title reads: “Visma is Signing Dopers to Defeat Pogačar…”

Thankfully, this channel doesn’t use Kirby’s voice, but Kirby is aware that accusations could be “levelled at me because of something a machine has said.”

Kirby isn’t taking any action at the minute, legal or otherwise, although word on the street is that British broadcaster and biologist David Attenborough is getting very animated about the use of his voice without having given his permission.

Instead, Kirby is choosing to look on the bright side. “My tones may live on forever; I could be calling the first Martian Tour de France.”

So, who is the person behind this Watts YouTube channel?

Luckily, buried within the YouTube channel is an email address. I send an inquiry that is eventually answered by a Slovenian called Max.

“I’m working on this channel together with a close friend of mine,” he begins in answer to a series of questions I’ve sent over. “I’d still like to stay somewhat anonymous, since it wasn’t ever my plan to publicly expose myself, but we are both huge cycling fans. We’re both employed and do this as a hobby, purely because we both enjoy it, and enjoy everything about cycling.”

Both Max and his friend had run different YouTube channels in the past, before noticing that beyond posting clips and highlights, no one was covering cycling in this news-y documentary style that started to become very popular in other sports a few years ago.

“I’m really glad you asked this,” Max says when asked to explain how he makes the videos, “and I really hope all those people who keep commenting that the videos are made completely using AI are reading this (lol).”

In short, Max finds the topics and writes the scripts before making the voiceovers (using the Carlton Kirby voice he’s generated). His friend does the video editing before handing it back to Max to upload it to YouTube with the thumbnail image that he has made, which will hopefully attract people to click on their creation.

“Many people tend to think you can create such videos using AI in seconds, but I’m here to tell you it’s impossible,” Max continues. “No AI can edit the videos this way, we’re not there yet. The only AI we use is for the voiceover, and that’s it.”

How did they end up using an AI voiceover that sounds like Carlton Kirby?

“We knew we had to use a British voice, simply because in our opinion, British commentators are one of the best, and really do fit the sport. We thought about using our own voice at first, but since we’re not native English speakers, we decided to go with a realistic AI voice,” Max says.

“It may be hard to believe, but it wasn’t actually our intent to mimic Carlton’s voice, at least not at first,” Max tells me. “We just used a random UK male model, tweaked it and kind of played with it for a while. We then started to notice similarities with Carlton’s voice, so we decided to go more towards that direction, since we really do look up to him and see him as one of the best sports commentators to ever exist. We still don’t think the voice we use is exactly the same as Carlton’s, but we do admit it’s fairly similar. So yeah, we decided to go with that voice, because to us, it just sounded epic. We felt like it kind of resonates with cycling and completely fits in, that’s why we did it.”

While he is glad that Kirby is deciding to see the use of a voice very much like his own as an homage, Max says he can also understand the commentator’s apprehension.

“I do agree having someone use a voice similar to yours to read stories can be a bit scary, especially if it’s something bad or unethical,” Max says. “In our case, we don’t do anything bad, and don’t use it to put words into someone’s mouth, therefore we don’t really see too big of an issue with it, and we hope he doesn’t either. Our sole purpose is to make the sport more respected and recognisable, and we really do believe our YouTube videos help with that. Being able to use a voice that sounds familiar to the voice of one of the sport’s best commentators is sheer privilege, and we’re happy to be able to do it.”

A measured response, with some merit, and they do lean away from the more potentially defamatory topics such as doping that others farm for clicks.

“So, to answer the question, yes, we have considered that,” he says of weighing up whether to use a voice that sounds like Kirby’s. “However, we do believe that since our intent is good, it isn’t too big of an issue. We never claimed Carlton himself is narrating any of our videos, and we don’t think that would be the right thing to do either, because he’s not. We’re not trying to impersonate anyone, just spread the word about the sport, that’s literally it. We’ve also got a note about it in our channel’s ‘About’ page, where it explains in detail why we use an AI voice.” 

This about page, which could also be interpreted as a disclaimer, reads: “The voice used in our videos is made with the help of AI technology, using a UK male model that we think best suits the sport. It’s important to note that even though the voice may sound similar to a real person, it is indeed an AI-generated voice, carefully crafted to sound as natural and pleasant as possible. It is by no means a voice meant to imitate anyone, put words in someone’s mouth, or violate anyone’s opinions or privacy. It is solely made for entertainment purposes, to tell the given story alongside the visuals on the screen.”

In a blog post published last year, YouTube offered its own thoughts on AI-created content and announced the ability to request that AI-generated content that simulates an identifiable individual be removed from the platform.

“We’ve heard continuous feedback from our community, including creators, viewers, and artists, about the ways in which emerging technologies could impact them. This is especially true in cases where someone’s face or voice could be digitally generated without their permission or to misrepresent their points of view,” the post reads.

“So in the coming months, we’ll make it possible to request the removal of AI-generated or other synthetic or altered content that simulates an identifiable individual, including their face or voice, using our privacy request process. Not all content will be removed from YouTube, and we’ll consider a variety of factors when evaluating these requests. This could include whether the content is parody or satire, whether the person making the request can be uniquely identified, or whether it features a public official or well-known individual, in which case there may be a higher bar.”

Right of privacy is one issue, but the right to control one’s own intellectual property for commercial use is another, even when – as in the case of Max’s Watts channel – there’s no explicit reference to a celebrity. Earlier this year, the actress Scarlett Johansson threatened legal action against OpenAI (the company behind ChatGPT) after it introduced an AI assistant that strongly resembled her voice. Although OpenAI never named Johansson, it agreed to stop using the voice.

Max, polite and eloquent throughout, finishes his reply with a broader rebuff of those who may criticise AI-created content. “I think it’s perfectly fine and nothing too serious, unless you’re trying to imitate and impersonate someone with an intent of scamming people or doing something else along those lines, that isn’t really ethical or for a good cause,” he says. “Telling someone it’s actually the person who said something, while it was actually AI, is wrong. Just using a voice that mimics someone’s voice is okay, as long as it’s clear you’re just using it for narration. But that’s just my opinion. As long as you’re not using it for anything bad, I really don’t think it’s a big deal.

“It’s a double-edged sword, of course, but in our case, it’s for the promotion of the sport of cycling, and we don’t see anything bad in it. We’ve never had anyone truly complain about the voice sounding too similar, so we hope they don’t mind it either.”

It would be easy to dismiss Max, to view this as a black and white issue that he’s fallen on the wrong side of. And yet, given how easy it is to generate a fake voiceover of anyone, including Carlton Kirby, maybe we should be a bit grateful that Max has kept things above board with his creations.

Did we do a good job with this story?