Resemble AI alternative: Resemble stopped selling voice cloning in 2026, so where do you clone your voice now

Written by Inam. VoiceClone, the last column in the table, is my own app, so keep in mind that I am not a neutral judge here. You can see my other engineering work on designesh.com. Last updated 26 October 2026.

You might have used Resemble AI before to clone your voice, or perhaps you came looking for it after reading a recommendation from 2025. When you open their website today, you will notice something is wrong right away. There is no price table and no sign up button on their voice cloning page, and the page simply says the product is no longer available.

This sudden change is why older lists of alternatives do not make sense anymore. Resemble AI no longer sells voice cloning to new customers, so if you are starting today you need a different tool. The company has shifted its focus to deepfake detection, and Chatterbox, its open model, is the part of their voice software that you can still use for free.

What happened to Resemble AI's voice cloning?

Resemble AI used to be one of the main cloud voice cloners people talked about. You sent them an audio sample, their machines built the voice model on their own computers, and you paid each time you wanted to generate speech. In fact, as late as May 2026, their changelog entry on voice cloning pricing said your first clone was free and every new clone after that was $2.

Then they changed their business completely. On 13 August 2026, their CEO, Zohaib Ahmed, published an article titled why a former voice AI company went all in on deepfake detection. He wrote that Resemble is supporting existing customers but not taking new ones, and they will not build new voice models to sell anymore.

If you look at the Resemble voice cloning page right now, it says in plain words that they have "stopped offering voice generation as a commercial product." You can check their pricing page as well, but it only shows plans for finding deepfakes, adding watermarks, and verifying identity, so voice cloning is not on any of them.

What does this mean if you already have a Resemble voice?

The CEO wrote that existing customers are still supported, so your account is not simply switched off. But he gave no final cutoff date, and he did not say anything about letting people download their trained models. A cloud voice clone never sits on your hard drive, because it is a model on someone else's server, and you can only talk to it through their website or an API, which is the connection between your software and their servers.

I suggest you look at this remaining time as a window to get your work done. You should generate every spoken line you know you need, save the sound files to your disk, and store your original voice clips safely, because you will need those exact recordings to build your voice again in another app. I saw the same thing happen when PlayHT shut down, and as I explained in what to do after PlayHT, the practical steps are the same.

Is Chatterbox a free Resemble AI alternative?

You can still run some of their voice code yourself, but only in a technical way. Chatterbox is an open text to speech model built by Resemble, and you can find the Chatterbox repository on GitHub published under the MIT licence, which means you are allowed to use it in paid work. The CEO wrote that it will stay alive under their research work, and it remains open and free for anyone to build on.

There are several versions, and you need to know which one fits your machine. Chatterbox-Turbo uses 350 million parameters and speaks English, Chatterbox-Nano uses 110 million parameters and is made to run on a standard processor, which the README notes runs about 3 times faster than real time on 8 cores, and Multilingual V3 uses 500 million parameters and covers 23 languages. None of them needs a training step because they copy a speaker from a short audio clip, and the README uses a 10 second clip to demonstrate it.

Chatterbox is free and you can sell what you make with it, but it is a Python tool, not an app you double click. Resemble says they built and tested the software on Python 3.11 using Debian 11, which is a version of Linux, so you have to install it with pip and call it from your own code. It also puts Resemble's Perth watermark into every audio file, which is a hidden signal that humans cannot hear. I went through these details, along with the setup errors people hit on Windows, in Chatterbox TTS in plain words.

Why do cloud voice cloners keep leaving people stranded?

This is a recurring problem with cloud services, and Resemble is the second case I have written about here after PlayHT. When you use a cloud voice cloner, you leave your voice on their machines and pay them to use it. That arrangement works fine while they sell the product, but it stops the day the owners decide to sell something else.

Open models solve the monthly subscription problem, but they create a technical hurdle instead. You do not pay any money for the software, but you have to install Python, set up your graphics card drivers, and repair the setup when a library updates. A software engineer might not mind spending an afternoon on that, but a creator who only wanted a voiceover for a video pays for cloud tools simply to avoid that headache.

What is the safer way to keep a voice?

The fix is simple: your voice should end up as files on your own disk, and the software should run without asking an outside server for permission to speak. When you run your tools locally, no company decision can take the voice away, because they never hold your files in the first place.

Keeping things local also protects your original audio. A voice recording is personal data that works almost like a fingerprint, and the safest way to keep it private is to never upload it. You keep your source clips, you keep your trained voice, and you can copy both to a second drive whenever you like.

How does VoiceClone do it, and where is Chatterbox better?

I built VoiceClone because paying rent for my own voice never sat right with me. It is a desktop program for Windows 10 and 11 that runs the open source GPT-SoVITS engine, and you launch it by double clicking one file named START VoiceClone.bat. The program downloads the engine on its first start, which is about 1.3 GB, and after that it runs fully offline and nothing is ever uploaded.

To set it up, you provide 1 to 3 minutes of natural talking, let the app check your recording, and press the Make it perfect button to train the voice on your own PC. The training is fast if you have an NVIDIA graphics card, and it still gets there on a normal processor, just slower. The free version lets you clone and fully train one voice with 3 generations to check it, and the $49 one time license opens unlimited length, the word editor, retraining and exports, with nothing monthly.

I want to be fair about where Chatterbox is stronger. Its Multilingual V3 model works with 23 languages while VoiceClone only handles English, its Turbo model understands tags like [laugh] and [cough], and it copies a voice from a short clip without any training step. If you need many languages and you are happy in Python, use Chatterbox. If you want your own English voice in a Windows app with no monthly bill, VoiceClone is built for that. You can download it from the download page, and I ask that you only clone your own voice or a voice whose owner gave you clear permission.

FAQ

Is Resemble AI shutting down?

No, the company is still in business, but it has moved its focus to deepfake detection. Its voice cloning page states that voice generation is no longer offered as a commercial product, and the CEO wrote on 13 August 2026 that it supports existing customers but is not taking new ones.

Can I still sign up for Resemble AI voice cloning?

You cannot join as a new customer today. The pricing page only lists plans for detection, watermarking and identity checks, so anyone wanting to make a new voice clone has to use a different tool.

Is Chatterbox free for commercial use?

Yes, you can use it in commercial work. The Chatterbox repository on GitHub is under the MIT licence, and you should know that every sound file it makes carries Resemble's Perth watermark, which you cannot hear.

What is the easiest Resemble AI alternative on Windows?

A local program saves you from both the cloud and the Python setup. VoiceClone works on Windows 10 and 11, clones and trains one voice for free, and the full English app is a single payment of $49.

Sources: the Resemble AI voice cloning page, pricing page and changelog, the post by Zohaib Ahmed on why Resemble went all in on deepfake detection dated 13 August 2026, and the resemble-ai chatterbox README and licence on GitHub, all read on 26 October 2026, plus the VoiceClone product page on myvoiceclones.com.