FakeYou Review 2026: I Made Morgan Freeman Read My Pizza Order. Then Things Got Weird.

My friend Marcus sent me a voice message at 2 AM on a Saturday. Morgan Freeman's voice, unmistakable, that velvet baritone narrating the Shawshank redemption of his stomach — "I would like one large pepperoni pizza with extra cheese, garlic knots on the side, and a two-liter Sprite, because hope is a good thing, maybe the best of things, and no good thing ever dies hungry." I listened to it four times. I sent it to my group chat. My roommate thought it was a voice actor doing a bit. It wasn't. Marcus had typed that text into FakeYou, picked the Morgan Freeman model, hit generate, and waited. That last part matters more than you'd think.

FakeYou is a free web tool at fakeyou.com where you type text and it generates audio in the voice of a celebrity, cartoon character, video game protagonist, or basically anyone the community has trained a model on. Morgan Freeman. SpongeBob. GLaDOS from Portal. Various anime characters I had to Google. The voice library is enormous and chaotic — part jukebox, part fever dream — because the models are community-uploaded. Anyone with the technical chops can train a voice model and add it to the platform. This means the selection is vast. It also means the quality is all over the place like a garage sale where some tables have antique silverware and others have broken clock radios.

I spent an entire Sunday afternoon testing it. Not because I had to. Because once you start making Barack Obama narrate your grocery list, stopping feels like leaving a party too early.

The way it works is dead simple. You go to fakeyou.com, pick a voice from the dropdown, type your text into the box, and hit the generate button. Then you wait. And wait. On a free account, the queue times are the defining experience of FakeYou. I timed them. My first generation, a Morgan Freeman clip, took about seven minutes. A SpongeBob one took eleven. A lesser-known anime character voice — Gojo from Jujutsu Kaisen, which my younger cousin insisted I try — took fourteen minutes because the queue was packed. There's a progress bar that moves with the confidence of a DMV line. You sit there watching it crawl from 0% to 12% to 12% to 12% to suddenly 67% and then it stalls at 89% for two minutes before finally finishing. The free tier experience is an exercise in patience.

And then the audio plays.

Some of it is genuinely impressive. The top-rated Morgan Freeman model sounds close enough that if you played it from another room, someone might do a double-take. The pacing is right. The cadence is right. That particular way he lingers on certain syllables. It's uncanny for about ten seconds, and then you notice the edges. A slightly metallic quality. Words that run into each other where a human would pause. The occasional syllable that sounds like it was assembled from spare parts, which it was. But for a meme? For a joke in a Discord server? For making your friend laugh at 2 AM? It works.

Other models sound like Morgan Freeman having an allergic reaction underwater. That's the thing about community-trained models. The quality depends entirely on who trained it, what data they used, how much of it they had, and whether they knew what they were doing. I tried maybe thirty different voices that Sunday. I'd say five were genuinely good, ten were recognizable but rough, and fifteen were bad enough that I wouldn't have guessed who they were supposed to be without the label. One Darth Vader model sounded more like a guy doing Darth Vader at a birthday party than like Darth Vader. Another one — actually wait, I need to be fair here. The Darth Vader model labeled "Darth Vader (Iconic)" was pretty solid. There are often multiple models for the same character, uploaded by different community members, and the quality gap between the best and worst version of the same voice can be staggering.

My friend Jen tried it and her reaction was immediate. "Why does Optimus Prime sound drunk?" She'd picked a low-rated model. I showed her how to sort by popularity and filter by the highest-rated versions. "Oh okay this SpongeBob one is actually scary good." Exactly. The platform has gems buried under a lot of mediocre uploads, and learning to navigate that is half the skill of using FakeYou.

So is FakeYou safe? This comes up a lot and I get why. You're visiting a website, typing text into a box, and downloading audio files. The site itself is legitimate. It's been around since 2021, it's operated by a company called Storyteller AI, and the site runs on a standard web stack. I didn't encounter any malware, sketchy redirects, or suspicious download prompts. Your browser is fine. Your computer is fine.

But "safe" has layers.

The community-uploaded model ecosystem means you're trusting that the audio you generate is just audio and nothing more. And it is — the output is a WAV or MP3 file, nothing executable, nothing hidden. But the ethical dimension is where "safe" gets complicated, and I think it's the most interesting thing about FakeYou. Because what you're doing, when you generate a Morgan Freeman clip, is using someone's voice without their permission. Morgan Freeman didn't consent to narrating your pizza order. The voice model was trained on recordings of him from movies, interviews, audiobooks. His voice, his identity, repurposed by strangers on the internet for laughs.

For memes and jokes among friends? Most people shrug at that. It feels harmless. Silly, even.

But the same technology that makes your group chat laugh can put words in someone's mouth that they never said and would never say. And FakeYou doesn't have meaningful guardrails against that. You can type almost anything and generate it in a real person's voice. I'm not going to pretend I tested the absolute worst-case scenarios, but I did notice that the platform's content moderation is minimal. There are terms of service. There are some restrictions. But the gap between what the TOS prohibits and what the tool actually prevents you from doing is wide enough to drive a truck through.

I don't think FakeYou is evil. I think it's a toy that exists in a legal and ethical gray zone that most of its users don't think about because they're too busy making Squidward read the Communist Manifesto. And honestly, that tension — between genuinely funny and genuinely concerning — is what makes it such an interesting product to think about.

Pricing. FakeYou has a free tier that lets you generate as many clips as you want, but you sit in a queue behind everyone else. Wait times range from five to fifteen minutes depending on traffic. The premium plans — FakeYou Plus starts at around $7 per month — give you priority queue access, which cuts wait times to under a minute. There's a higher tier for even faster processing and some API access. If you're using FakeYou regularly, the premium is almost mandatory because the free queue times will make you question your life choices. Seven dollars a month for a meme tool is a personal decision. I paid for one month, used it heavily for a week, made every famous voice say increasingly unhinged things, and then canceled because the novelty wore off. Your trajectory may vary.

The audio quality caps out at a certain point regardless of what you pay. FakeYou isn't competing with professional voice synthesis. The models are community-trained, the inference runs on shared infrastructure, and the output is usually 22kHz mono audio. Fine for phone speakers and Discord. Not fine for a YouTube video or a podcast or anything where someone might listen on decent headphones and notice the artifacts. This is a toy. A spectacular, weird, occasionally jaw-dropping toy. But a toy.

So what if you want something that isn't a toy?

ElevenLabs is the obvious comparison, except it's not really a comparison. ElevenLabs is a professional voice synthesis platform where the voices sound human enough to make you uncomfortable. (If narrated presentations are more your speed, Narakeet is another AI voice tool worth knowing about.) Their voice cloning feature lets you upload samples of any voice and create a synthetic version that is, frankly, indistinguishable from the real thing in short clips. The quality ceiling is miles above FakeYou. But ElevenLabs costs real money — plans start around $5 per month for limited characters and go up to $99 and beyond for serious usage. And ElevenLabs takes the ethics question more seriously, with voice verification requirements for cloning. You can't just clone Morgan Freeman because you feel like it. You need to prove you have rights to the voice, or you're using one of their pre-made voices. Different product, different rules, different price point.

Uberduck occupies similar territory to FakeYou — community voice models, meme-friendly, pop culture characters — but with a slightly different model library and interface. If you tried FakeYou and couldn't find the specific voice you wanted, Uberduck might have it, and vice versa. The quality variation is comparable. The wait times are comparable. It's like choosing between two thrift stores on the same block. Same vibe, different inventory.

And then there's a completely different category of TTS that people sometimes land on while searching for FakeYou, which is read-aloud tools. People who Google "fakeyou" aren't always looking for celebrity voice memes. Some of them just want to hear text spoken aloud — articles, documentation, study material, long emails from their landlord — and they stumbled into the celebrity voice corner of the internet by accident. If that's you, what you actually want is a reading tool, not a voice generation toy.

CastReader is ours, so take this with the appropriate grain of salt, but it's a Chrome extension that reads web pages aloud with paragraph-by-paragraph highlighting as it goes. No celebrity voices. No meme potential. No waiting in a queue for twelve minutes to hear SpongeBob say something unhinged. Just clean, natural-sounding TTS that reads the article you're looking at while you cook dinner or fold laundry or pretend to exercise. Completely different use case. I mention it because the overlap in search intent is real — I've seen people in our support emails say they tried FakeYou first and realized it wasn't what they needed. If you want to listen to a 3,000-word blog post while doing dishes, a browser extension that highlights where you are in the text is a better tool than a website that generates fifteen-second clips of cartoon characters.

Back to FakeYou. Here's what I keep coming back to.

It's fun. Genuinely, uncomplicatedly fun, in a way that a lot of AI tools aren't. Most AI products want to be taken seriously. They want to be your copilot, your assistant, your second brain. FakeYou wants to make Homer Simpson say something about your ex. And there's a purity to that. The Discord servers and Reddit threads full of FakeYou clips are some of the funniest corners of the internet, people doing absurd creative things with a tool that has no pretensions about what it is.

But I can't review it without saying plainly that making someone's voice say things they didn't say is a power that deserves more guardrails than FakeYou currently provides. The fun doesn't cancel out the risk. Both things sit in the same room, and using the tool means accepting that.

If you want to try it, go to fakeyou.com, pick a popular model with high ratings, type something short and funny, and be prepared to wait. Sort by top-rated to avoid the garbage models. Don't pay for premium unless you've already used the free tier enough to know you'll keep coming back. And maybe think for a second before you generate that clip about whose voice it actually belongs to.

Marcus still sends me FakeYou clips. Last week it was David Attenborough narrating his cat knocking a glass off the counter. I laughed for a full minute. Then I thought about how David Attenborough might feel about it, and I laughed a little less. That's FakeYou in two sentences.