Clone your voice on your own PC: private voice cloning for Windows
Record about 15 to 20 seconds of speech and Mimicore builds a voice you can use live, in text to speech and in video dubbing. The cloning runs on your computer, so your recording never goes to our servers.
Voice cloning without handing your voice to a website
Most voice cloning tools work the same way: you upload a recording to someone's server, wait, and download the result. That is convenient, but it also means your voice, one of the most personal pieces of data you have, now lives on a company's infrastructure under that company's terms. Mimicore takes the opposite approach. Cloning, the live voice, text to speech and dubbing all run on your own PC, and the voice you create is saved in a folder on your own disk.
To be precise about "offline": the voice work itself happens locally, but the app does need the internet for a few ordinary things. You sign in with Google, Discord or an email code, the app downloads its models the first time, and it checks your plan. If you go offline later, your plan stays valid for up to seven days, so a trip or a bad connection does not lock you out. What never travels is the sound of your voice.
Two kinds of cloned voice
Mimicore offers two routes to a voice of your own, and they suit different goals. Instant cloning needs about 15 to 20 seconds of natural speech and is ready in under a minute. It is the fast way to hear yourself in another language, to get a voice for narration, or to try an idea. Instant voices work live (with about half a second of delay at the best quality), in text to speech across 23 languages, and in dubbing.
Trained voices use RVC, a technique that learns one specific voice from 10 to 30 minutes of clean recordings. The result is a closer likeness with latency depending on the Latency and quality slider position and your PC, and it is the better choice when you want a voice that holds up over long sessions. You can train on your own NVIDIA graphics card or, if you do not have one, for free in the cloud on a Google Colab GPU in your own Google account. Our page on RVC for beginners walks through that route step by step.
What makes a good recording
The quality of a clone is limited by the recording it starts from, so a little care here pays off more than any setting afterwards. Choose a quiet room and speak close to the microphone at a normal, steady volume. Read naturally and continuously, the way you would explain something to a friend, rather than reading a list of words. If the recording is too quiet Mimicore tells you, and if it clips you should move back a little. Mimicore removes background noise automatically, but a quiet room still sounds better than a noisy one that has been cleaned up.
Use a decent microphone if you have one, but do not worry if you only have a headset: a close, clean headset recording clones better than a distant expensive microphone in an echoey room. Avoid music, other voices and heavy reverb in the sample, and speak in the language and energy you want the voice to carry, because the clone inherits some of that character.
Clone your voice in four steps
- 1
Open Voices & effects and press + Instant voice. Give the voice a name and choose whose voice it is: your own, or someone who gave you permission.
- 2
Record about 15 to 20 seconds of natural speech in a quiet room, or import a clean recording you already have.
- 3
Press Create voice. It is ready in under a minute and stays on your PC.
- 4
Use it: pick it on Live and press Start, type text on Speak, or dub a video in your own voice.
What you get
Private by default
Cloning, the live voice and speech generation run locally. Cloud training is optional and uses your own Google account only.
23 languages
A cloned voice can speak Arabic, German, English, Spanish, French, Hindi, Japanese, Korean, Portuguese, Russian, Chinese and 12 more.
Works live
Use your clone as a live voice in Discord, OBS, Zoom or games through the virtual microphone.
Mix voices
Blend two of your voices with a slider to make a new character that belongs to no real person.
Calibrate to your voice
Read one sentence and Mimicore sets the pitch and noise gate for the closest match to your microphone and room.
Your files, your folder
Voices live in %LOCALAPPDATA%\Mimicore\voices. Copy the folder to move them to another PC.
The Free plan includes one instant voice of your own. Unlimited cloning, mixing, importing and training are part of the Studio plan, and new accounts start with 30 free Studio days that need no card.
Consent and responsible use
Cloning a voice is powerful, and it comes with responsibility. Clone your own voice, or the voice of someone who has clearly agreed to it. Do not clone a public figure, an ex, a colleague or a stranger without permission, and do not use a cloned voice to deceive, defraud, harass or impersonate. This is written into Mimicore's acceptable use policy, and there is a way to report misuse if you see it.
The voice gallery shows the safe alternative: its 19 voices are original blends of two speakers, so none copies a real person, and they are free to use. If you want a distinctive voice without cloning anyone, start there or mix two of your own voices.
What you can do with a cloned voice
The obvious use is narration: type a script and have it read in your own voice instead of recording take after take. Creators use this to fix a flubbed line without re-recording the whole video, to produce voice-overs in languages they do not speak, and to keep a consistent voice across a long project. See text to speech in your own voice for how the Speak page works.
A second use is live: stream, game or call as a character that you built, with the same voice every time, instead of tweaking pitch sliders. A third is dubbing: translate a video into another language and hear it in your own voice, with audio and subtitle export. Because all of this runs on your PC, there are no per-minute fees for generating audio, and no queue.
What your PC needs
Windows 10 or 11 (64-bit). Instant cloning and live use work best with a DirectX 12 graphics card from NVIDIA, AMD or Intel, and Mimicore picks the best card automatically, falling back to the CPU if there is none. Expect the instant voice add-on to take roughly 1 to 2 GB of disk space, depending on the model size chosen for your PC. Training on your own PC needs an NVIDIA GTX 16 or RTX 20 series card or newer with at least 4 GB of memory and about 10 GB of free disk space, while cloud training works on any computer.
If the download looks unfamiliar to Windows, you may see a SmartScreen warning, because Mimicore is new and not yet code-signed. Choose More info and then Run anyway. The installer is small, installs for your Windows user only and needs no admin rights (only the optional virtual-mic setup asks for admin once), and every update is signed and verified by the app before it installs.
Questions
Does Mimicore upload my voice recording?
No. Instant cloning runs on your PC and the voice is saved on your disk. Only if you choose optional cloud training do recordings go anywhere, and then only to your own Google Drive and Colab, never to Ribanix Labs servers.
How long a recording do I need to clone my voice?
About 15 to 20 seconds of natural, continuous speech is enough for an instant voice. A trained RVC voice needs 10 to 30 minutes of clean recordings of one person.
Does it work completely offline?
The voice work runs locally, but you need internet to sign in, to download models and to check your plan. If you go offline, your plan stays valid for up to 7 days.
Can I clone someone else's voice?
Only with their permission. Clone your own voice or one you have clear consent to use, and never use a clone to deceive or impersonate. See the acceptable use policy.
Is voice cloning free?
Yes, you can clone one voice of your own for free, forever. Unlimited cloning, mixing and training are part of the Studio plan, and new accounts start with 30 free Studio days without entering a card. The voice gallery, 28 effects and the soundboard are free forever, and your own voice works live for 60 minutes a day on the Free plan.
Can a cloned voice speak other languages?
Yes, in 23 languages. A clone created from your English speech can read text in Spanish, Hindi, Japanese or any other supported language, though accents and quality vary by language.
More: Voice cloning and RVC training · RVC for beginners · Text to speech in your own voice · AI video dubbing · Full user guide · Help and FAQ