Stop Paying for ElevenLabs: Free AI Voices with Claude
A walkthrough of installing the open-source VoiceStudio with Claude Code, then testing voice cloning, voice design and dubbing against ElevenLabs.
Posted
yesterday
Duration
Format
Tutorial
educational
Views
11K
310 likes
57 · 43
Big Idea
The argument in one line.
An open-source voice studio running locally, installed by Claude Code with no manual coding, can replace paid ElevenLabs credits for most everyday voiceover, cloning and dubbing work.
Who This Is For
Read if. Skip if.
READ IF YOU ARE…
A YouTuber or podcaster who pays for ElevenLabs credits every month and wants to know whether a free local tool can cover routine voiceovers.
A non-developer who wants to see Claude Code handle a full GitHub install, including dependency and memory errors, without typing commands by hand.
A freelancer weighing voiceover, audiobook or video dubbing services and wanting a zero-per-generation cost way to produce them.
SKIP IF…
You need the most expressive voices or real-time conversational audio, where the video itself says ElevenLabs is still the better pick.
You want a rigorous quality benchmark; the comparison here is a single informal A/B listening test, not a measured scorecard.
TL;DR
The full version, fast.
The video claims a free, open-source tool called VoiceStudio can stand in for ElevenLabs, which charges up to $99 a month and meters usage in credits. The host opens a terminal, creates a folder, launches Claude Code, and pastes one prompt asking it to download the project, install dependencies, set up a local cloning model and launch the app. Claude resolves dependency and memory errors, the host picks the recommended OmniVoice model, and the app runs locally. He then clones his voice from a 10-second sample, designs a late-night radio voice from a text description, and covers audiobooks and dubbing. His verdict: not perfect every time, but free per generation, and check the model license before paid client work.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
ElevenLabs plans run from a starter tier up to $99 a month, and even paid users hit credits that run out.
A local voice tool removes per-generation cost, so you can experiment with voices without watching a credit meter.
Claude Code can install an open-source project from a GitHub repo, diagnose dependency and memory errors, and launch the app with one pasted prompt.
A 10-second voice sample recorded in a quiet room was enough for the cloned output to sound close to the original recording.
Voice design lets you describe a voice in words, such as a deep calm late-night radio host, instead of recording anything.
Running locally means recordings stay on your computer and supported features work offline after models download.
Free does not mean equal: the host still recommends ElevenLabs for expressive voiceovers and real-time conversation.
Model licenses differ, and some voice models restrict commercial use, so check the license before using output for paid client work.
Voiceovers for YouTubers, audiobooks for authors and dubbing for creators are services people already pay for, and zero marginal voice cost improves the margin.
Cloning a voice should only be done with the speaker's permission.
The video uses a blind A/B listening test between a paid ElevenLabs line and the free tool, with the answer withheld until the end.
Takeaway
Free local voice tools are viable, with caveats
WHAT TO LEARN
A local open-source voice app removes per-generation cost, and a coding agent removes the install barrier, but quality, licensing and consent still need your judgment.
Paid voice tools meter usage in credits, so cost scales with how much you generate; local models trade that for hardware time and a one-time setup.
Letting Claude Code install an open-source repo works because it can read errors, fix dependency and memory problems, and relaunch without you knowing the commands.
Choosing the recommended model pack during setup is the practical default; heavier models cost more download size and memory.
A short, clean 10-second sample in a quiet room is enough input for a convincing clone, so recording quality matters more than length.
Voice design from a text description is useful when you need a distinct character voice and have no reference audio.
Judge free against paid with a blind A/B test on your own script, not on the vendor's demo lines.
Check each model's license before selling output, since some local models limit commercial use.
Only clone voices you have permission to use, and keep recordings local when privacy matters.
Voiceovers, audiobooks and dubbing are existing paid services, and removing per-generation cost raises the margin on them.
Glossary
Terms worth knowing.
VoiceStudio
An open-source desktop app for voice cloning, voice design, audiobooks, dubbing and transcription that runs on your own computer.
OmniVoice
The recommended local voice cloning model chosen during VoiceStudio setup. It downloads onto your machine so generation needs no credits.
Voice cloning
Generating new speech in a specific person's voice from a short recording of them speaking.
Voice design
Creating a brand new synthetic voice from a text description of how it should sound, with no reference recording.
AI dubbing
Translating a video's speech into another language and generating new audio for it.
Claude Code
Anthropic's terminal-based coding agent, used here to install and configure software from a plain-language prompt.
The Script
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
17px
Right now, AI voice is moving faster than almost anything else in AI. 11 Labs recently launched 11v4. You can clone a voice, design a completely new one, dub videos into different languages, and do some pretty incredible things with AI audio.
But all of that comes with a price. Starter, creator, pro, all the way up to $99 a month. And even after paying, you're still dealing with credits that run out.
But here's the question. What if you could do all of this without paying anything for voice generation? Before I show you how, let's try something.
You're about to hear the same line twice. One was generated using a paid 11Labs plan, and the other was generated using a completely free tool running on a normal laptop. Hey, I'm Matus, and I'm here to make AI simple, useful, and completely free of the usual hype.
No unnecessary buzzwords. No complicated jargon. Just...
Real AI tools, honest reviews, and things you can actually use. Hey, I'm Midas, and I'm here to make AI simple, useful, and completely free of the usual hype. No unnecessary buzzwords, no complicated jargon.
Just real AI tools, honest reviews, and things you can actually use. Now, which one do you think was free? Comment A or B below.
I'll reveal the answer at the very end of this video. And here's where things get interesting. This free tool can clone your voice, create completely new voices from a description, and even dub videos into different languages.
It runs locally on your computer, so there's no monthly subscription or credits running out every time you generate something. And we're going one step further. We're connecting it with Claude.
So Claude won't just write a script for you. It can actually turn that script into audio using your own voice. But wait.
I've saved the biggest surprise for the end. Because what if I told you that this free tool could do more than just save you money? What if you could actually use it to start earning money?
And I'm not talking about some complicated AI business. I'm talking about services people are already paying for. I'll show you exactly how towards the end, so make sure you don't miss that part.
For now, here's what we're going to do. First, we'll install this free tool using Claude code, without manually writing any code. Then we'll clone our own voice, design a new one, try dubbing a video, and see how it compares with Eleven Labs.
And yes, we'll find out which voice was free in that comparison. Every source, the full scorecard, and a whole library of AI tools and resources are on midas .ai. Links in the description.
Alright, let's get started!
But before we get into the actual setup, if you've ever used Eleven Labs to create voiceovers, you know how easy it is. You type your script, generate the audio, and you're done. But if you're creating videos regularly, those credits can run out pretty quickly.
Well, that's where Voice Studio comes in. It's an open -source alternative to Eleven Labs that lets you clone your own voice, design new voices, dub videos, transcribe recordings, and even create entire audiobooks, with support for up to 646 languages across its features. Now, think about it.
Voice cloning, AI voiceovers, video dubbing, these are features you'd normally have to spend so many dollars for in Eleven Labs, especially if you're using them regularly. But here, you're getting similar tools for free. Seriously, no monthly voice credits, no paid API keys, and you don't have to worry about wasting credits every time you want to try something new.
You can experiment with different voices, create as many voiceovers as your computer can handle, and actually explore these tools without constantly thinking about the cost. And since everything runs locally, your recordings can stay on your computer. You can even use supported features offline once everything is downloaded.
On top of that, you get dictation, automated workflows, and integrations with AI agents, all in one place. Now, imagine how useful this could be if you're creating YouTube videos, podcasts, or voiceovers every single day. That's a lot of money you could potentially save.
Of course, free doesn't automatically mean it's as good as Eleven Labs. And that's exactly what I want to find out. So, in this video, I'm going to put it to the test.
We'll set it up using Cloud Code, explore what it can do, and see how closely it can clone my own voice. And since the whole idea is to make this easier, let's start by letting Cloud Code handle the installation. Installing Voice Studio with Cloud Code.
Now that we know what Voice Studio can do, let's get it running on our computer. And don't worry, you don't need to be a developer for this. We're going to use Cloud Code to handle most of the technical work, so you can follow along without having to figure out every command yourself.
Step 1. Open the GitHub repository. First, open Google and search for Voice Studio.
You'll find its GitHub repository right here. Open it, and you'll see all the project files, supported platforms, and installation instructions. I'll leave the GitHub link in the description so you can find it easily.
Step 2. Create a project folder. Now, instead of installing everything manually from here, we're going to let Claudecode do the heavy lifting.
So, open Terminal on your computer. First, type this command. mkdir dash p tilde slash voice dash studio to create a folder called voice dash studio.
Then, type cd tilde slash voice dash studio. This simply takes us inside the folder where we'll set everything up. Step 3.
Launch clod code. Once we're inside the folder, type clod and hit enter. This will launch clod code right inside our project folder.
If it asks you to trust the folder, just confirm it, provided it's the folder you created. And now, instead of writing a bunch of installation commands ourselves, we can simply tell clod what we want. Step 4.
Give clod the installation prompt. I'm going to paste this prompt asking Claude to download Voice Studio, install all the required dependencies, set up a free local voice cloning model, and launch the application. I'll also ask it to fix any errors along the way and make sure the voice generation and audio export are working properly.
And that's pretty much all we need to do for now. Let's see how Claude handles the rest. Step 5.
Let Claude install everything. So, Claude starts by downloading the Voice Studio project from GitHub. Once that's done, it checks our computer, installs the required packages, and starts setting up the local AI engine.
Now, this might take a little while because we're not just installing a regular app. We're also setting up the AI models that will generate the voices. And during our installation, we actually ran into a few issues with dependencies and memory.
But instead of trying to solve everything manually, we let Claude investigate the errors and work through them. Step six, choose the voice cloning model. Now, Claude is asking us which voice cloning model we want to install.
You can see a couple of options here, but I'm going with Omnivoice, which is the recommended one. And now, the model starts downloading onto our computer. This is an important part, because once it's installed, we can use the supported local voice generation features without paying for generation credits.
Step 7, launch Voice Studio. All right, once the installation is ready, we'll ask Claude to launch Voice Studio. The app opens with a quick system check to make sure everything is working.
Then we choose the fast model pack, let it install, and go through the remaining privacy and permission settings. And there we go. Voice Studio is finally running on our computer.
So we've gone from a GitHub repository to having our own local AI Voice Studio set up, with Claude handling most of the installation. Now comes the interesting part. We've got the software running, but how good are the voices it actually creates?
Let's find out. All right, so Voice Studio is finally running on our computer. Now let's see what we can actually do with it.
And as you can see, the interface is pretty simple. We've got different sections for voices, stories, dubbing, and tools. Let's start with the one I'm most excited about, voice cloning.
I'm going to record a quick 10 -second sample of my voice in a quiet room and upload it here. Then I'll select a local voice cloning model, enter a completely new sentence, and hit generate. All right, let's hear what it sounds like.
Here's a fun fact. Octopuses have three hearts, blue blood, and nine brains. And when they swim, one of their hearts actually stops beating.
Now let's compare that with my original recording. And look at that! It's actually so similar to my original voice!
And we just generated that on our own computer, without spending anything on voice credits. Now, if you're not completely happy with the result, you can try another recording or experiment with a different model. And once you're satisfied, you can export the audio and use it in your projects.
Just remember to clone voices only with permission. Now, there's also voice design, where you can create a completely new AI voice. Let's try describing a deep, calm voice, like a late -night radio host, and generate it.
Good evening, America. You're listening to 98 .7 FM, midnight radio, broadcasting live from New York City. I'm your host, bringing you smooth sounds, great stories, and a little company through the night.
Stay tuned. That's pretty interesting. We just created a new voice without recording anyone.
and the Voice Gallery has different voices and languages to explore. Moving on to Stories. Here, you can turn written stories into narrated audio, create storybooks, and even produce full audiobooks.
And the cool part? Voice Studio can automatically analyze your story, identify male and female characters, and cast the most suitable voice for each. Pretty cool, right?
And this is exactly the kind of thing we love exploring at Matus. Finding useful AI tools, testing what they can actually do, and showing you how to use them without spending a fortune. Now let's check out dubbing.
I'm going to upload a short English audio. Select any language and generate a dubbed version. Hello everyone!
Welcome back! Today, we're exploring how AI is changing the way we create content. From generating realistic voices to translating videos into different languages, the possibilities are pretty exciting.
Let's check it out! And there we go! We've got the same audio in a completely different language.
Of course, the voice might not sound exactly the same, but for voiceovers, podcasts, and other audio content, this could be really useful. There's also batch dubbing for working on multiple videos and transcribe, which turns audio into text. You also have a profile section where you can manage your settings and preferences.
And one feature I really like is A -B voice comparison. You can listen to two versions side by side and see which one sounds better. Apart from these, you've also got directorial AI, speech rate fit, probe file, audio conversion, and a few other tools to explore.
Now, remember how we used Claude code to install Voice Studio? Well, we can actually connect Claude to Voice Studio too. Well, not exactly.
Eleven Labs is still more convenient if you want to generate voices directly in your browser. And it can be better for expressive voiceovers and real -time conversations. But if you're creating regular voiceovers, audiobooks, or dubbed content, Voice Studio gives you a free alternative worth trying.
Just check the model's license before using its output for paid client work. And for these kinds of tasks, Voice Studio can replace the need for an Eleven Labs subscription without paying for every generation. Now, remember how I told you that you could actually make money using these tools?
Well, here's how. You can actually get paid for AI voiceovers for YouTubers, create audiobooks or audio storybooks for authors, or dub videos for creators who want to reach a global audience. And the best part?
These are services people already pay for. But with Voice Studio, you don't have to spend money on every voice generation. So think about it.
You're not just getting a free alternative to 11 Labs. You're getting a tool that could help you start earning money, too. We started with one goal.
to find a free alternative to Eleven Labs that is free. And we actually made it work. We installed Voice Studio using cloud code, cloned our voice, generated voiceovers, and explored everything from audiobooks to video dubbing.
And the results? Pretty good. Sure, it's not perfect every time, but the fact that we can do all of this locally without paying for voice credits is what makes it so exciting.
And as we saw, you could even use these tools to get paid services and potentially earn money. Now, if you want to try it yourself, the complete setup guide, the exact cloud code prompt, and all the links are in the description. At Midas, we find and test tools like this so you can use powerful AI without expensive subscriptions.
So subscribe if you want more free AI tools and workflows. And remember, the next time you think you need to pay for an AI tool, check if there's a free alternative first. You might be surprised by what you find.
The Hook
The bait, then the rug-pull.
ElevenLabs charges up to $99 a month and still runs out of credits. The host plays two identical lines, one paid and one from a free local tool, and makes you guess which is which before showing how Claude Code installs the free one.
Add Modern Creator as a preferred source and Google shows you more of our breakdowns in Search, Top Stories, and AI Overviews. It only changes what you see, and you can undo it in your Google settings anytime.
Add to Preferred SourcesOpens your Google source preferences with us pre-loaded. Tick the box and you're done.
Same terrible coffee shop site, same one-line prompt, 100 rounds each for Sonnet 5.5 and Opus 5.5. Sonnet finished at $22 and under 3,000 lines. Opus ran 8.5 hours, cost $137, and buried the site in features.
Andrew Warner and Adam walk through the week's trending open-source AI repos, argue over where copying software becomes theft, and finish with five viewer-submitted first-star projects.
A walkthrough of Claude Code Projects, the cloud orchestrator that runs a fleet of persistent Claude sessions on your subscription, with a full demo build.
A walkthrough of the 11 biggest changes to the Claude desktop app, from the session rail and master sessions to artifacts, mods, and keyboard shortcuts.
Riley Brown argues the loudest AI story is backwards: the people with the most agency are using it to delete slop, from TV bloatware to app upsells to bloated SaaS.
00:00
00:17
00:29
00:33
00:46
00:57
01:07
01:19
01:27
01:38
01:45
01:55
02:12
02:17
02:30
02:42
02:46
03:01
03:06
03:19
03:31
03:39
03:50
04:03
04:10
04:21
04:33
04:42
04:55
05:05
05:12
05:29
05:37
05:48
06:02
06:09
06:18
06:28
06:39
06:47
06:55
07:10
07:22
07:33
07:37
07:52
08:00
08:07
08:22
08:37
08:40
08:54
09:04
09:13
09:25
09:35
09:44
09:56
10:06
10:12
10:23
10:38
10:46
10:58
11:06
11:15
11:26
11:43
11:50
12:02
12:11
12:21
12:32
12:42
12:52
12:58
13:08
13:24
13:29
13:44
13:50
33:59
18:49
17:06
28:05
14:06