A short story about voice cloning, low expectations, and unexpected money.
First of all – no, I did not deserve to get my account banned.
I used to do voice work on Fiverr. VeRRrrRy bAAaaAAdlLLyyy as one “professional agency” once told me – but my customers loved it. I had recurring customers, from all over the world, 5,0 stars (Yes, I did!!) and lots of fun recording.
I loved doing voice work, but it’s a pretty gatekept industry.
If you haven’t been to film school, studied theater, or attended some other kind of arts university, there’s very little room for career changers who want to break into voice work. On top of that, voice work isn’t what you think it is.
It’s non glamorous and annoying. You spend 12 hours in dark rooms filled with spiky foam and moody directors, because apparently nobody cares about working hours.
It’s also underpaid, and there’s no fixed income, so you might end up in a shared flat with other artists who are either on drugs, deep into method acting, or out of work. None of that appealed to me, so even though I loved doing voice work, I never wanted to become an established voice actress. I just loved doing voices. It’s a beautiful art that can make anything better.
I made ads for well-known brands, and ads for small businesses and startups. Explainer videos. A small video for a lawn mower manufacturer whose name I still struggle to pronounce. I was decent at it, the gigs came in steadily, my income eventually reached the low four figures, and I genuinely enjoyed the whole thing.
And then my account got banned.
Not for anything I did wrong.
My oldest son, a teenager, entrepreneurial spirit, wanted to make some money, created his own Fiverr account from our shared home network. Same household.
Second account. Fiverr’s system flagged it, and both accounts were gone. No appeal. No second chance. Just a big fuck youuuuu from fiverr.
So there I was.
Wannabe-voice-actress, plattformless and slightly bitter about the whole thing.
I started googling. platforms for voice over 2025? And over several YouTube videos and Reddit posts, I found ElevenLabs.
Not the voice cloning for content creation thing – I tried that before and it was unusable in my language. The marketplace. The part where you sell your cloned voice as a licensable asset and earn royalties every time someone uses it.
I made my first voice, put it in the store, and didn’t touch it again for a week. I mean, because there’s nothing to touch anyway.
I made €15!!!!
No marketing, no promotion. No me doing anything at all. The platform just… did it while I was sleeping, and doing other shit.
That was enough to make me take this seriously.
What the ElevenLabs Voice Library actually is
ElevenLabs has a searchable marketplace of AI voice clones. Developers, content creators, companies building voice agents, they browse it, pick a voice they like, and use it to generat speech. Every time they do, the voice creator (that’s you, if you’re in the library) gets paid.
It’s a licensing model, not a gig model. You’re not trading time for money like I did on Fiverr.
You record once, upload, and then the royalties happen in the background while you do something else.
The default payout rate is around $0.03 per 1,000 characters ; roughly 90 seconds of spoken audio. That doesn’t sound like much. But a single voice that gets picked up for a conversational AI application, or a product demo, or an audiobook project, can generate thousands or millions of characters a month.
ElevenLabs has paid out over $5 million to voice creators through this system so far. The platform is valued at $11 billion and growing fast. The demand for voices – especially for Conversational AI – is not slowing down.
I currently have seven voices in the marketplace. They perform very differently. Some sit quietly and earn at least more than they cost. One consistently earns, the others are supporting actors.
Let’s talk about Conversational AI
Conversational AI is the use case right now.
Companies are building AI phone agents, customer support bots, virtual assistants. All of them need voices. Not just any voice – natural-sounding, warm, real-sounding voices that don’t make people want to hang up immediately.
Think about what that means for what you record.
If you have a good telephone voice – clear, natural, friendly, not overly performative, that’s your angle. Record like you’re actually on the phone. Not like you’re in a voiceover booth.
How to actually do this
What you need to start
- An ElevenLabs Creator Plan ($22/month , get 50% off with my link for the first month, AND you can downgrade after uploading; your voice stays live and keeps earning)
- A decent microphone (USB works fine, clarity matters more than expensive equipment)
- A quiet space (closet with clothes is honestly great for this)
- 2-3 hours of patience
What to record and how
ElevenLabs requires a minimum of 30 minutes of audio for a Professional Voice Clone, but they recommend 2-3 hours for significantly better quality.
Most important part for Conversational AI voices: Record in-character for the use case you’re targeting.
For a telephone/agent voice, record how that character actually sounds:
- “Hello, this is [name], how can I help you today?”
- Natural follow-up questions
- Explanations, confirmations, polite redirects
- Some warmth, some professionalism, a bit of real-life hesitation (ehhm, ahh etc)
Not scripted-sounding. Not voiceover-booth-smooth. Actual conversation. The model learns from what you give it, and it will sound like whatever you recorded.
For a narration voice, read. Books, articles, long-form content. Vary your pace. Express something. Two hours of genuinely expressive reading produces a much better clone than two hours of flat recitation.
Technical checklist:
- Record in a quiet room with minimal echo (soft surfaces help)
- Consistent distance from mic throughout
- No background noise, no street noise
- Clean audio, trim silences at the beginning and end of files
- Consistent energy, don’t start tired and end caffeinated
The upload process
- In your ElevenLabs account, go to Voices → Add New Voice → Professional Voice Clone.
- You’ll complete a consent verification step so ElevenLabs can confirm it’s actually your voice and that you have permission to clone it.
- Upload your audio files. ElevenLabs recommends at least 30 minutes of clean spoken audio, and says results are usually better with closer to 2+ hours.
- Wait for processing — the clone quality depends heavily on what you uploaded. Clean, consistent recordings with minimal background noise make a major difference.
- Test it thoroughly before publishing.
Getting your voice into the marketplace
Once your clone is ready, go to your voice settings and make it shared and discoverable so it can be listed in the Voice Library. ElevenLabs does a brief review before it becomes searchable.
Tags matter enormously. Be specific: accent, age range, tone, language, and use case all help people find your voice. A clear description usually performs better than something vague and generic.
Don’t write “versatile voice for any project.” Write something like: “natural-sounding mid-30s British accent, optimized for conversational AI and customer service applications.” That matches how buyers actually search.
Upload a 10–30 second preview clip that shows your voice at its best.
Apply for the HQ badge if your audio quality is high. HQ voices can charge higher rates — up to about €0.17 per 1,000 characters, compared with the default rate of about €0.03 per 1,000 characters. These are approximate EUR conversions from ElevenLabs’ published USD payout figures. I never got into the HQ badge thing, because I never recorded enough hours I guess. If I work it out, then there’ll be another guide 🙂
How you get your money
Payouts go through Stripe Connect and are typically processed weekly, or roughly every 6–8 days, once you hit the minimum payout threshold. ElevenLabs states that threshold is usually $10, which is about €8.58 at the June 1, 2026 exchange rate.
The default rate of about €0.03 per 1,000 characters means one 10-minute YouTube script earns roughly €0.23. On its own, that’s not exciting — but the upside is scale. If your voice gets picked up by a popular app or workflow, the same script format can be used thousands of times.
My first week: €15. Without any marketing or extra effort.
Some of my voices have earned more since. Some are still quiet. What I’ve noticed is that voices recorded with a clear use case in mind tend to perform better than more generic ones. Conversational AI voices seem to outperform broader, less specific positioning. The general logic behind that fits how the Voice Library is structured around searchable voice attributes and use cases.
Whatever the users need can be tagged, found, and used.
The part I want to bothers me though
Yes, this affects professional voice actors. I know. I wanted to be one.
The creative industry – film, TV, entertainment, dubbing – is genuinely struggling with this.
People are losing work that they built careers around. I’m not going to pretend that doesn’t matter or wrap it in a cheerful “but disruption creates new opportunities!” bow.
I think about it.
A lot..
I pay for Apple Music. I buy albums. I buy films. I love concerts. Hell, I might even go to the cinema. I do what I can as a consumer to support the creative ecosystem I care about. But, what is music going to be like in 20 years? And the movies… How will writers survive all of that?
But with my all my hard thinking and rooting for the industry, I can’t stop Disney from firing a thousand people because their executives decided that derivative AI output is cheaper than human creativity. This is how the industry I am rooting for reacts to the market situation, and my not participating in the ElevenLabs marketplace doesn’t reverse it.
What I can say honestly: voice agents are as unavoidable as the internet was. Not good or bad as a statement, just a fact about where things are going. The question isn’t whether this technology should exist. It’s what you do with the skills and the voice you have, given that it does. I’m just as frustrated as every artist I’ve spoken to about this, but the industry is going to change for good. All of us need to figure out what part we play in that change.
For me: I cloned my voice, put it on a platform, and earned passive income that replaced something I’d lost. That felt like agency in a situation where I had very little of it.
If you’re thinking about trying this
You don’t need to have been a professional. You need a clear voice, a quiet room, and a few hours.
If you’ve ever been told you have a good phone voice, a nice reading voice, a distinctive accent – that’s enough to start.
→ Try ElevenLabs 50% off in the first month
Start with the free tier to explore, then upgrade to Creator when you’re ready to publish to the marketplace. You can downgrade again after, to the smallest plan, the voice stays live.
And if you do try it: come back and tell me what happened. I’m genuinely curious what voices end up doing well, and I’m still figuring this out myself.
It’s already in the afternoon today and I’m having a beer – cheers!
See you next Sunday. 🎙
Affiliate note: The ElevenLabs link above is an affiliate link, I get a small commission if you sign up through it,
no extra cost to you. I’ve been using ElevenLabs for a while and the voice marketplace income is real. I wouldn’t write a whole guide about something I hadn’t actually done.