So I decided to narrate my own audiobook with a free voice cloning tool. Here's what actually happened.

Last month, I finished writing a 200-page fantasy novel. My friends kept saying, "You should make an audiobook!" But when I checked professional narration rates, I nearly choked — $200–$400 per finished hour. My book would run about 8 hours. That's $1,600–$3,200. For a guy who writes in his spare time, that's not happening.

So I did what any broke creative would do: I Googled "free voice cloning" and fell down a rabbit hole. I found a site called Web666 that promised free voice cloning, online, no sign-up, no credit card. My first thought: this is either a scam or it's going to sound like a robot from 1998.

I was wrong on both counts. But it wasn't all smooth sailing either. Let me walk you through the whole messy, wonderful, occasionally infuriating process — because if you're thinking about doing this, you need the real story, not the marketing hype.

The setup: 30 seconds of my voice, zero dollars

The tool's voice cloning page asks for a sample of your voice. Just a few seconds, it says. I recorded a 40-second clip on my phone — reading a random paragraph from a news article, trying to sound natural but not overacting. Uploaded it, clicked a button, and waited.

Processing took about two minutes. When it finished, I typed a test sentence: "The dragon circled the castle, its scales glinting in the moonlight." The playback nearly made me jump. It sounded... like me. Not a perfect clone, but a very close cousin. The intonation was right, the slight rasp in my voice was there. I'm not gonna lie, I got a little emotional. My own voice, reading my own words, for free.

But then I tried a longer paragraph. And that's where things got interesting.

Chapter 1: The first 3,000 words — and my first frustration

I pasted the first chapter of my novel into the text box. It's about 3,000 words. The tool processes text in chunks, so I had to split it into several batches. Each batch took maybe 15–20 seconds to generate audio. Not bad, honestly.

But here's the thing — the free version has a character limit per generation. I hit it on the second batch. I had to manually cut my chapter into smaller pieces)Skip — that was annoying. There's no auto-split feature. I spent ten minutes just chopping paragraphs.

Still, once I got the rhythm down, the audio quality was surprisingly consistent. No robotic pauses, no weird emphasis on the wrong syllables. The tool even handles punctuation well — it pauses at commas and periods like a human would.

One thing that threw me off: the free voice cloning AI sometimes adds a slight breath sound at the start of a sentence. It's subtle, but once you notice it, you can't unhear it. I ended up editing those out in my audio software, which took extra time.

The speed test: 20 chapters in three evenings

Here's the part that surprised me the most. I was expecting this to take weeks. But after the first chapter, I got into a flow. Each chapter (about 4,000–5,000 words) took roughly 45 minutes to process, including my manual splitting and the generation time.

I did the math: 20 chapters × 45 minutes = 15 hours. Spread over three evenings, I had a full audiobook draft. Compare that to recording myself — I tried that once, and it took me four hours to record one chapter because I kept stumbling over words. The free voice cloning software was about 5× faster.

And the best part? No retakes. If I didn't like a sentence, I just regenerated that batch. No studio, no microphone, no "take 47" frustration.

The quality check: where it shines and where it flops

I'm not going to pretend it's perfect. Let me break down what I found, because you need to know the real deal.

The good:

  • Dialogue sounds natural. My characters have different voices — the grizzled old wizard, the sassy teenage thief. The tool didn't give them distinct voices, but it handled the emotional tone well. Anger, excitement, sarcasm — all came through.
  • Pronunciation of fantasy names? Surprisingly solid. I made up words like "Vaelithor" and "Krynnath" — it nailed them on the first try. I was expecting to have to add phonetic spellings, but I didn't.
  • Long-form narration stays consistent. No drift in pitch or speed over 20 chapters.

The not-so-good:

  • The free voice cloning tool sometimes struggles with very long sentences. If I had a 50-word sentence with multiple clauses, it would occasionally lose the thread and flatten the intonation. I learned to break those up.
  • It can't do character voices. Every character sounds like me. For a single-narrator audiobook, that's fine. But if you're writing dialogue-heavy scenes with distinct voices, you'll need to edit or accept the sameness.
  • Emotional extremes are hit-or-miss. A scream? It comes out as a slightly louder version of my normal voice. Not great.

But here's the thing — for a free tool, this is remarkable. I compared it to a paid TTS service I used last year for a work project. The paid one cost $30/month and sounded slightly more human, but not $30/month better. For my audiobook, the free version was good enough.

The money question: did I actually save hundreds?

Yes. Let's do the math.

If I had hired a narrator at the low end ($200 per finished hour), my 8-hour audiobook would cost $1,600. With the free voice cloning tool, I paid $0. The only costs were my time and the electricity to run my computer.

But I should be honest about the hidden costs. I spent about 10 hours editing the audio — removing breath sounds, fixing a few mispronunciations, and adjusting pacing. If you value your time at $20/hour, that's $200 of my labor. Still, I came out $1,400 ahead.

And the tool's free voice cloning text to speech page has a bunch of advanced settings — pitch, speed, emphasis. I played with those for a while)Skip — they help, but they're not a substitute for a human director.

What I wish I knew before I started

If you're thinking about trying this, here's my honest advice:

  1. Record a clean sample. I recorded mine in my living room with the TV on in the background. Big mistake. The tool picked up a faint hum that I didn't notice until I listened with headphones. Re-record in a quiet room, close to the mic.
  1. Break your text into small chunks. The free version has a character limit. Plan for that. I'd say 1,000–1,500 characters per batch is the sweet spot. It gives you enough context for natural pacing without hitting the limit.
  1. Listen to every batch. Don't just generate and move on. I found a few errors that I had to fix by regenerating. It's tedious, but it's the difference between a decent audiobook and a professional one.
  1. Use the advanced settings sparingly. The pitch and speed controls are tempting, but I found that leaving them at default gave the most natural result. When I tried to make my voice sound deeper, it started to sound like a bad impression of a movie trailer guy.
  1. Have a backup plan. The free tool is free forever, but it's an online service. If the site goes down, you're stuck. I saved all my generated audio files locally, which I recommend.

The verdict: should you try it?

If you're a writer with a finished manuscript and no budget for narration, yes. Absolutely. The free voice cloning tool at Web666 is legit — no sign-up, no credit card, no hidden trial period. It's not going to replace a professional voice actor for a commercial audiobook, but for your own project, for a gift for your mom, for a podcast intro, it's more than enough.

My audiobook is now live on my website. I'm not making money from it, but I don't care. I just wanted to hear my story read in my own voice, and I did. That feeling is worth more than the $1,600 I didn't spend.

So go ahead. Record a few seconds of your voice, paste in a chapter, and listen. You might surprise yourself. I know I did.

P.S. If you try it, come back and tell me how it went. I'm curious if your experience matches mine — especially the breath sound thing. Maybe it's just me.