Help and FAQ

Mimicore is new and not yet code-signed, so Windows SmartScreen warns about it. Click More info, then Run anyway. The installer installs for your Windows user only and needs no admin rights (only the optional virtual-mic setup asks Windows for admin once). Every update is signed and the app verifies the signature before installing it.

Windows 10 or 11 (64-bit). A DirectX 12 graphics card from NVIDIA, AMD or Intel is recommended for live voices; the 28 effects run on any PC. Training your own voice on the PC needs an NVIDIA GTX 16 or RTX 20 series card or newer with 4 GB or more of memory and the driver installed (and at least 10 GB of free disk space). No NVIDIA card? Train free in the cloud instead.

Open the Support page and press Check my setup. It runs quick, read-only checks that finish in a few seconds: Windows version and WebView2, whether Windows blocks desktop apps from the microphone, the microphone (it listens for about a second and shows the level; a completely silent microphone is usually muted in Windows or by a button on the device), headphones or speakers, the virtual cable (VB-CABLE), the graphics card with its video memory and driver age, the add-ons (file sizes and SHA-256 against the list that ships with the app; a button checks every file), free disk space, whether GitHub (updates) and the Mimicore server (sign-in) answer, whether a newer version exists, and your plan. Each line is OK, Note, Check or Fix, with a button for the fix where there is one. Nothing is changed, downloaded or started, and no internet is fine for Live. Copy report copies a text without file paths, e-mail addresses, your Windows user name or PC name; Send with feedback opens Send feedback with that report in the message.

Yes: the page mimicore.pages.dev/requirements (and the System requirements box on the download section of the home page) reads what your browser reports (Windows version, memory, processor count and the graphics card name) and says in plain words what should work. It runs in your browser, sends nothing and makes no network request. A browser can only guess: it cannot read your video memory or the exact Windows 10 version, and some browsers hide the graphics card. After installing, the app's own Check my setup is the authoritative answer. On a phone or tablet the page only says that Mimicore is for Windows PCs.

Windows 10 and 11 (64-bit) is the main version. There is also a free Mac beta for Apple Silicon Macs (M1 or newer, macOS 12 or later), with every feature unlocked while it is in beta: download it from the "Mac beta" section at the bottom of the Mimicore website home page. There is no Linux, Android or iPhone version.

Download the Mac beta from the website. The beta is not notarized by Apple yet, so the first time open it with right-click, then Open. To send your voice to Discord, Zoom, OBS or a game, install the free BlackHole 2ch audio driver (the Live page walks you through it), then choose BlackHole 2ch as the microphone in the other app. Live voice changing needs a fast Mac; the effects, Speak, Dub and cloud training run on any Apple Silicon Mac. The Mac beta sends diagnostic logs (no audio, no typed text) to help fix problems; switch this off in Settings.

The installer is about 22 MB. Guided setup asks what you want to do (Live, voiceover or Dub) and downloads only what that needs, for example Instant voice cloning (about 1 to 2 GB, the model size is picked for your PC) and/or Trained voices (RVC) (0.3 GB). No background performance test runs. Optional add-ons: Trained voices 0.3 GB, Voice Training 2.5 GB (NVIDIA only), Video dubbing about 3.2 GB, the voice gallery 27 MB.

You do not have to choose: the Instant voice cloning add-on installs the tier that suits your PC (Fast for weak or integrated graphics, Balanced for mid-range cards, Ultra for strong cards with 6 GB or more) and tunes it. Manual tiers are under Settings > Add-ons > Advanced: model tiers. For reference: Balanced (about 1.3 GB) is the recommended choice for most PCs. Fast (1.1 GB) has the lowest delay and graphics memory use, best for gaming and low-spec PCs. Ultra (1.8 GB) has the most polished speech and files but is about 3 times slower. Every tier speaks 23 languages and you can add or remove tiers any time in Settings > Add-ons.

No admin rights: it installs for your Windows user only. The live voice, effects and speech run on your PC, 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, and free live minutes you use offline are sent the next time you are online.

Yes. Sign in with Google, Discord or an email code (Settings > Account, or when the app asks). It is free: new accounts in India get 30 days of Studio, and outside India the Early Bird Gift gives Studio free while international payments are being set up, then the Free plan stays free forever (28 effects, the soundboard, unlimited live voice with the built-in and gallery voices and 60 minutes a day with your own voices).

In Mimicore choose Settings > Account > Sign in, pick Google, Discord or email, and confirm on the website page that opens; the app then signs in by itself. If you use email, the 6-digit code can take a minute: check spam, and ask for a new code if it is older than a few minutes. If a sign-in link seems stuck, close the website tab, press Sign in in the app again and use the new page. Still stuck? Create a support ticket in Support.

Mimicore checks for updates when it starts and every few hours, downloads them in the background and asks you to restart. You can also check yourself: Settings > About > Updates > Check for updates. Updates are signed and verified before installing.

If you reported a problem through Support, you get an email with a link to an early build containing the fix, so you can test it before everyone else. It is never installed automatically: open Settings > About > Updates > Check for early release. Afterwards press It is fixed or Still broken on your ticket in the Support tab.

Uninstall like any app: Windows Settings > Apps > Installed apps > Mimicore > Uninstall. Uninstalling does not delete your data folder (%LOCALAPPDATA%\Mimicore), so your voices and settings stay. Delete that folder yourself to remove everything. If you moved the add-ons to another folder (Settings > Storage), the uninstaller does not delete that folder either: delete it yourself.

Yes. Settings > Storage > Where the add-ons are stored > Move to another folder…, then pick a folder on a drive that is built into this PC (not a network drive, USB stick or card). Mimicore creates a Mimicore folder there, copies the add-on models (about 2 GB) and the Voice Training pack (several GB), checks every file, and asks you to restart; after the restart it uses the new place and deletes the old copy. Cancel at any time leaves everything as it was. Voices, outputs and projects stay in the data folder. Settings > Storage also shows how much space each part uses and cleans up temporary files, leftovers of stopped downloads and old results safely (voices, recordings, unfinished trainings and saved projects are never touched).

The drive that holds your add-ons is not connected (or its letter changed). Reconnect it and restart Mimicore, or in Settings > Storage choose another folder (pick the folder that contains the Mimicore folder) or press Use the default folder (the add-ons then have to be downloaded again). Downloads stay paused until you decide so that nothing is downloaded twice by mistake.

The folder must be on a drive built into this PC and must not be a network location, a removable drive, a cloud-synced folder (OneDrive), a link or junction to somewhere else, the Windows or Program Files folder, or Mimicore's own data folder. It must be writable and have enough free space for the move. The Microsoft Store version also cannot use folders inside AppData.

In %LOCALAPPDATA%\Mimicore. Voices are in the voices folder, generated speech in outputs, logs in logs (mimicore.log), settings in settings.json. Open it from Settings > Privacy & data > Open data folder.

Install Mimicore on the new PC and sign in. Copy the voices folder from %LOCALAPPDATA%\Mimicore\voices on the old PC to the same place on the new one to bring your voices. You can use Mimicore on all your PCs (1 active PC at a time).

Open Live > Output > Set up virtual mic. Mimicore downloads the free VB-CABLE driver from vb-audio.com, checks its signature and installs it (click Yes when Windows asks; restart if Windows asks). Then in the other app choose CABLE Output (VB-Audio Virtual Cable) as the microphone. Discord: User Settings > Voice & Video > Input Device. OBS: add an Audio Input Capture source and select CABLE Output. Zoom, Teams, Meet: Settings > Audio > Microphone. Games: set the in-game voice input device, or set CABLE Output as the Windows default recording device.

In Discord User Settings > Voice & Video turn off Discord's own Noise Suppression (Krisp) and Echo Cancellation. They fight the changed voice. Keep Mimicore's Noise suppression on instead.

Check that the voice is started (the big Start button shows Live), that Live > Output > Virtual mic is not Off, and that the other app's microphone is CABLE Output (VB-Audio Virtual Cable), not your normal microphone. If the cable does not show in the other app, restart Windows after the VB-CABLE install and open Mimicore again.

Restart Windows after the VB-CABLE install, then open Mimicore again. Restart the other app too (Discord, OBS and games only list audio devices when they start). If it still does not appear, run Live > Output > Set up virtual mic again.

Click Yes on the Windows permission prompt, close other audio apps, and try again. If Windows asks for a restart, restart and press Set up virtual mic once more. Mimicore downloads the VB-CABLE package from vb-audio.com and checks its signature; a blocked download (a firewall, VPN or antivirus) is the usual cause. If it keeps failing, create a support ticket in Support so we can see the log.

Use headphones, or turn off Hear myself on the Live page. With speakers, the microphone picks up its own output.

Turn on Hear myself on the Live page (Output section) and pick your headphones as the output. Use headphones so the microphone does not pick up the speakers.

On Live, pick a voice or effect from the list and press the big Start button, then speak. The status line shows estimated local delay and glitches; the Live engine panel shows processing time. Press Start again (or use the start/stop hotkey) to stop.

Stop Live on a trained voice and try the slider toward Higher quality for more context and processing time. If the first seconds of a session glitch on a long position, that is the output buffer still settling; it is larger at the start of long positions and shrinks after about 20 seconds. The Lower latency end processes more often and may sound less natural. More delay does not repair a microphone recovery fault. Use wired audio, check the selected microphone and Settings > General > Processing. If no position works smoothly, try a lighter voice or free effect and export a testing log.

Mimicore 2.9 gives trained voices a Latency and quality slider inside the voice selection card, with 15 positions from about 80 ms to 1.2 s of buffering in 80 ms steps; keep it toward the left for calls and games. Left reduces buffering; right favours quality and adds delay. Stop Live before adjusting. Your choice stays fixed. The Live panel estimates local delay; lookahead, inference, audio devices, the chat app and the network all contribute to the real mouth-to-ear delay. No position guarantees performance on every PC. Effects have no slider.

Stop Live on a trained voice and move the slider toward Lower latency, then restart. Shorter buffering needs more frequent processing: if it causes gaps, compare a position toward Higher quality or use a lighter voice/free effect. Use wired audio and close graphics-heavy apps. Low GPU utilisation alone does not establish smooth processing. Overload drops stale input instead of increasing delay automatically.

Mimicore 2.9 replaces the old Live mode and Block size controls with one Latency and quality slider directly below the trained voice name. It has 15 fixed choices from about 80 ms to 1.2 s of buffering and hides buffer sizes. Stop Live before adjusting. Old long-buffer presets map to the nearest supported chat setting; voice, device and pitch data are preserved. Effects retain their own processing. GPU load depends on the voice and device.

Keep Automatic level on (quiet microphones get louder), or raise the mic level in Windows sound settings. If it is too loud or clipping, move back a little or lower the level in Windows. Calibrate to my voice shows warnings for quiet or clipping microphones.

Turn on Noise suppression and set the Noise gate to Medium (Off, Light, Medium or Strong). The noise gate silences everything between your words.

On Live, open the Microphone section and choose your microphone from the list (the default is Windows default). Press Refresh after plugging in a new one. The level meter moves while the voice is running. If you unplug a microphone, Mimicore waits for it or falls back to the default microphone.

Mute silences your microphone without stopping the voice, so unmuting is instant. Original-voice test (bypass) passes your real voice through unchanged so you can check levels.

For trained voices only. Pitch shifts in semitones (plus 12 is one octave up). Voice strength sets how strongly the sound is pulled toward the trained voice. Both apply instantly.

For trained voices. Read one sentence and Mimicore measures your pitch and your room, then sets the pitch and noise gate for the closest match. The settings are saved for that voice and load when you pick it.

On a fresh install Mimicore walks you through every page once. After an update that changes the interface it shows a short What's new tour with only the changes; press Escape or Skip tour to close either. Both appear only once.

In Mimicore 2.9 you choose latency and quality for trained voices with the slider on Live. There is no startup benchmark, background performance test or automatic quality adjustment. Model downloads are explicit, and selecting a graphics card is separate from the slider. Existing automatic profiles are discarded when upgrading; your choice stays fixed until you change it.

Processing time is the main indicator: processing must keep up with incoming audio. The panel also shows the slowest 1 in 20 blocks, app CPU share, busiest GPU engine, video memory, sound gaps and estimated local delay. A serial CPU stage, transfers or one GPU engine can miss the deadline while total GPU use is below 100%. These numbers do not measure the whole Discord or meeting call.

Mimicore 2.9 keeps your selected latency and quality position fixed. Stop Live to adjust the trained-voice slider yourself. Saved set-ups include the selection. Retired automatic-quality profiles do not adjust it.

Try a free voice effect first; effects need little processing. For a trained voice, start with the Latency and quality slider toward the left and compare a few positions to the right while checking the processing time against the block length. If it still cannot keep up, try Faster, slightly less accurate under the slider. Another voice may perform differently, but smooth real-time conversion is not guaranteed on every system. Plug laptops in, use Best performance and close graphics-heavy apps. Cloud training creates a voice; Live conversion still runs on this PC.

Average processing time is approaching or exceeding the block budget. This is a missed audio deadline, not a claim that the whole GPU is fully loaded. Mimicore keeps your slider position and bounds stale input; sustained overload can still cause audible gaps. Stop, move the slider to another position, try the Faster option or a lighter voice, and export a testing log if it persists.

Live: trained voices have one Latency and quality slider (15 positions, about 80 ms to 1.2 s of buffering) instead of Low, Medium and High, an optional "Faster, slightly less accurate" setting, a short notice when the graphics card is too busy, and a small background job that keeps a graphics card from slowing down between long blocks. Speak and Audiobook Studio: saved projects, PDF, Word and scanned-page import, chapters that resume after an interruption, a Speed slider and saved drafts. Voices: accent matching for trained voices, training that asks whether the voice is for speech or singing, and one training on this PC plus one in the cloud at the same time. Setup and safety: guided setup, Windows-protected sign-in and Drive details, and safer updates.

Yes. In Speak your script, title and voice settings are saved on this PC as you type, so closing Mimicore does not lose them. Each script is a project: you can rename, archive, restore and reopen it. Projects are stored in your Mimicore data folder and are not uploaded.

Yes. Audiobook Studio renders chapter by chapter and keeps a journal. When you open the same project again, finished chapters are checked and kept, and the render resumes with the next one (use the resume action on the project). If a finished chapter fails the check it is rendered again.

Yes, but only when you use it. In Audiobook Studio, "AI script polish" and the random narration paragraph send the script (or your instructions) through our support service to an AI provider (currently Groq) so a language model can write the result. This needs a signed-in account and a script of limited length; when you are signed out, the script is too long or the service is unavailable, Mimicore applies its built-in local cleanup instead and nothing is sent. Do not paste passwords or private data into a script you polish with AI. Everything else in Speak, including rendering the voice, stays on your PC.

A switch under the slider for trained voices, off by default; it never turns on by itself. It matches your voice against a smaller part of its training data, so each block finishes sooner on a slow or busy graphics card. The voice may sound a little less like the original. Switch it on before you press Start and off again any time; it applies the next time you press Start, and saved Live set-ups remember it.

Live noticed that several blocks in a row arrived late. Plug in a laptop's charger, close or limit games and apps that use the graphics card (for example by lowering a game's frame rate), or try a position closer to the left of the slider or the Faster option. The notice appears at most once in 15 minutes and goes away by itself; it never changes your settings.

Yes. Use Live > Add an effect and pick Radio, Cave, Robot or any of the 28 effects; it is layered on top of your voice.

Mimicore reopens an unplugged microphone, speaker or virtual cable by itself and shows a notice while it waits. Plug the device back in or choose another one; if the default microphone is available the live voice continues with it.

After an error in the voice engine Mimicore passes your own voice through so the call is not silent. Stop and start the live voice to reset it. If it happens again, create a support ticket in Support.

Voices & effects > + Instant voice, give it a name, choose whose voice it is (your own, or someone who gave you permission), record about 15 to 20 seconds of natural speech in a quiet room (or import a clean recording), then Create voice. It is ready in under a minute and works in Speak, Dub and file conversion. Instant voices work in Speak, Dub and file conversion; for Live add the Instant live voices add-on in Settings > Add-ons (it needs a DirectX 12 graphics card, and plays about 1 second behind you). Live lists your trained voices, and your instant voices once the add-on is installed. Background noise is removed automatically. The Free plan includes one instant voice of your own; creating more than one needs Studio.

Unlimited live use of your own voices, creating more than one instant voice, mixing, importing and training voices, and clips longer than the Free limits (3 minutes for Speak, 2 minutes for Dub and converting) are part of the Studio plan. You can start a 7-day Studio free trial on your account page (a refundable small payment-method check; billing starts after the trial; cancel any time before).

About 15 to 20 seconds of natural, continuous speech in a quiet room, close to the microphone, not too loud. If the recording is too quiet Mimicore says so; if it clips, move back a little. If there is lots of background noise Mimicore removes it, but a quieter room sounds better.

When you create or import a voice the app asks whose voice it is: your own, or someone else's with their permission. There are no spoken consent statements. The terms you accepted on the download page still apply: clone only your own voice or voices you have permission to use.

19 original ready-made voices, each a blend of two speakers so none copies a real person. In Voices & effects click Get the voice gallery (27 MB) once, press the play button to hear a voice and Add to my voices to use it. Gallery voices are free to use.

It blends two of your instant voices into a brand-new voice. Move the slider between A and B and press Create mix. Trained (RVC) voices cannot be mixed yet. Mixing is part of Studio.

Training copies a voice from 10 to 30 minutes of clean recordings of one person. On this PC (NVIDIA card needed): install the Voice Training add-on, add recordings, choose a quality and press Start training (Quick test about 10 to 20 minutes, Good about 25 to 50 minutes, Best about 1 to 2 hours). Without NVIDIA: choose Cloud (free), which trains on a free Google Colab GPU in your own Google account with the same trainer and quality. Training is part of Studio.

Training time depends on the quality you choose: Quick test about 10 to 20 minutes, Good (recommended) about 25 to 50 minutes, Best about 1 to 2 hours on your own NVIDIA card. In the cloud (free Google Colab GPU) it takes longer: about 25 to 40 minutes, 50 to 90 minutes and 1.5 to 3 hours. Keep Mimicore (or the Colab tab) open while it trains.

Training on the PC needs an NVIDIA graphics card (GTX 16 or RTX 20 series or newer, 4 GB or more) with its driver installed, and at least 10 GB of free disk space on the drive. Use Cloud (free) instead on any other PC.

Fixed in version 2.6.0: a PC with a usable NVIDIA card now stays on This PC. Update Mimicore (Settings > About > Updates). If you are on 2.6.0 or later and it still happens, create a support ticket in Support.

Voices & effects > Train a voice > Cloud (free): Connect Google Drive and allow access (Mimicore can see only the files it creates), add recordings, choose a quality and press Upload and open Colab. In the Colab tab press the play button, click Run anyway and allow Drive access. Keep the Colab tab open: Mimicore shows progress and installs the voice automatically. Your recordings go only to your own Google Drive and Colab, never to our servers, and are moved to your Drive trash afterwards unless you chose to keep them.

Free Google GPUs are limited at busy times. Try again later; training continues from the last checkpoint. Keep the Colab tab open while it trains. You can restart a stopped training by starting again with the same name.

Voices & effects > "Import voice..." opens one.mimivoice file (the single file that "Export voice..." in a voice's... menu creates). It shows the voice, checks and unpacks the file with live progress (you can cancel), asks you to confirm you have the right to use the voice, and installs it only if every check passes. A damaged or tampered file is refused and nothing is installed. If a voice with the same name exists you can keep both (default) or replace it. Older voice folders (voice.onnx, meta.json, optionally index.npy) still import through the "Older voice folder..." link in that dialog. Original PyTorch (.pth) weights are not supported - convert them with the training pack first. Only import voices you have the rights to use. Importing is part of Studio. A.mimivoice file can also be opened by double-clicking it in Windows (Open with Mimicore) or by dropping it on the Mimicore window; the same import dialog opens and nothing is imported without your confirmation. Files on a network share are refused: copy them to the PC first.

On the voice card: Rename, Delete (stop the live voice first), Preview to hear a sample, Convert a file to turn a recording into that voice.

On the Free plan live voice changing with the built-in and gallery voices is unlimited, and your own trained voices can be used live for 60 minutes a day (refills daily); the 28 voice effects and the soundboard are unlimited. Studio removes the daily limit for your own voices.

Partly. A trained (RVC) voice changes the sound of a voice, not how words are pronounced, so Live always keeps your own accent. In Speak, Dub and Convert you can switch on Accent for a trained voice: add a reference recording of that person (10 to 20 seconds of natural speech, kept only on your PC), then switch Accent on in Voices (open the voice's Accent section), in Speak or in Dub, and set the strength. The text is then spoken in the manner of the reference recording before the trained voice is applied, so part of the accent carries over. It is not an exact copy. Convert a file with accent matching needs the Dubbing add-on, because the words are recognised and spoken again. There is no live accent matching yet; an experimental mode with about 2 seconds of extra delay is planned.

Yes. In Speak, under Style, the Speed slider goes from 0.50 (half speed) to 2.00 (double speed) and the pitch of the voice stays the same. 1.00 is the normal speed. Dub keeps fitting each line to the timing of the original video, so it ignores this slider.

On the Live page, Voice tuning has the Pitch slider for the trained voice you are using, and "Calibrate to my voice" records about 10 seconds of you reading and sets the Live pitch for that voice; it is saved with the voice and loads whenever you pick the voice. Speak and Dub already aim for the voice's own pitch on every sentence, so most people only need the trim if the voice sounds too high or too low.

28 effects such as Robot, Demon, Chipmunk, Radio, Cave, Cathedral, 8-bit, Giant, Ghost, Elf, Dark Lord, Stadium announcer and Underwater. They run instantly on the CPU and are free with no time limit. Press the play button to hear one and Use to pick it for Live; Apply to a file processes a recording.

Soundboard has built-in sounds (air horn, applause, drum roll, sad trombone and more). Add sounds brings in your own MP3, WAV, OGG, FLAC or M4A clips up to 5 minutes, as many as you like. Click a tile to play it into the virtual microphone; turn on Also play on my speakers to hear it yourself. Give each sound a hotkey, set volumes in Edit and drag tiles to reorder. Sounds play even when the live voice is off.

Turn on Play into the virtual mic on the Soundboard page and make sure the other app uses CABLE Output as its microphone.

Open Speak, type or paste text (or load a.txt file), pick a voice and a language, adjust Emotion and Variation if you like, then choose Speak now (plays on your speakers and/or the virtual microphone, so people in your call hear it, and a copy is saved) or Save to file (renders a WAV).

23 languages: Arabic, Danish, German, Greek, English, Spanish, Finnish, French, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Dutch, Norwegian, Polish, Portuguese, Russian, Swedish, Swahili, Turkish and Chinese.

On the Free plan Speak clips are limited to about 3 minutes (about 2,700 characters), and Dub a video and converting audio files to clips of up to 2 minutes. Studio has no limits.

Playback starts after the first sentence (about 1 second); long texts are rendered sentence by sentence. Trained voices speak through the default voice, converted with the pitch matched automatically. The Ultra tier is the most polished for files but slower.

In %LOCALAPPDATA%\Mimicore\outputs, or where you choose with Save to file. Generated files are marked as AI-generated in their metadata.

Dub a video works on the Free plan for clips up to 2 minutes (no limit on Studio). Install the free Dubbing add-on once (Settings > Add-ons, about 3.2 GB). Open Dub a video, choose a file (MP4, MOV, MKV, M4A, MP3 or WAV), the spoken language (or Detect automatically), the target language and a voice, set Original audio, and press Dub it. Mimicore transcribes, translates sentence by sentence, speaks each line in your voice and fits it to the original timing, fully offline. You get a WAV audio track and an SRT subtitle file to add in any video editor.

Install it in Settings > Add-ons (Video dubbing, about 3.2 GB), then return to Dub a video. Installing the add-on is free; Dub a video is limited to 2-minute clips on the Free plan and unlimited on Studio.

Not yet: it creates a WAV track and an SRT subtitle file. Put the audio on the video in any editor (CapCut, Premiere, DaVinci Resolve, Clipchamp) and add the subtitles if you like.

Settings > Hotkeys: click a box, then press a key or combination. Available actions: start/stop the live voice, push-to-talk (hold), mute/unmute, original voice on/off, next/previous voice or effect, switch straight to a voice, and stop all soundboard sounds. Hotkeys work while a game has focus. Use combinations (Ctrl + Alt + key) or keys you do not use elsewhere (F7 to F12, numpad): a bound key is reserved for Mimicore while it runs.

Another app may already use that key, or the key is reserved. Pick a different combination such as Ctrl + Alt + a letter. Hotkey problems are listed in Settings > Hotkeys.

Free forever (after the 30 free Studio days): 28 voice effects, the soundboard, hotkeys, push-to-talk and the virtual microphone (all unlimited), unlimited live voice changing with the built-in and gallery voices, 60 minutes a day with your own voices, one instant voice of your own, Speak clips up to 3 minutes (about 2,700 characters), and Dub a video and converting audio files for clips up to 2 minutes. Studio: unlimited use of your own voices live, unlimited cloning, mixing, importing and training, no clip limits for Speak, Dub and converting, plus new premium features as they arrive.

Free forever (after your 30 free Studio days): 28 voice effects, the soundboard, hotkeys, push-to-talk and the virtual microphone (all unlimited), unlimited live voice changing with the built-in and gallery voices, 60 minutes a day with your own voices, one instant voice of your own, Speak clips up to 3 minutes (about 2,700 characters), and Dub a video and converting audio files for clips up to 2 minutes. Studio removes every limit: unlimited use of your own voices live, cloning, mixing, importing and training, and no clip limits for Speak, Dub and converting.

There are two plans: Free and Studio. Studio: unlimited use of your own voices live, unlimited cloning, mixing, importing and training, and no clip limits for Speak, Dub and converting. Studio has a low monthly price, a cheaper yearly price and a one-time lifetime price: the prices are shown in your own currency on the website account page (Choose your plan), and the yearly plan is the best value.

Yes. Every new account receives 30 days of full Studio access with no card needed (customers outside India currently get an Early Bird Gift of Studio access instead). Nothing is charged and nothing renews automatically; when the 30 days end your account continues on the Free plan forever and your voices stay on your PC.

Every account has an invite link and code on the website account page (Invite friends section) and in the app. Friends who join with your link get extra free Studio days. If you are on Free or temporary Studio, you earn 30 extra Studio days when they subscribe (up to 6 friends). If you already own Studio Lifetime or have an Early Bird Gift, your access is permanent and your link gives fellow creators free Studio days. Type a friend's code (FRND-...) in the code box on the account page and press Redeem within 14 days of signing up.

Yes. The app and the account page show a reminder in the last week when nothing renews your plan, and we notify you before onboarding or trial access ends. Subscriptions with autopay renew by themselves; cancel any time on the account page.

New accounts start with 30 days of Studio, no card needed (outside India: the Early Bird Gift). When subscribing to the monthly or annual Studio plan you can also start a 7-day trial where a small refundable amount (shown at checkout in your currency) verifies your payment method and sets up autopay, refunded immediately. You get full access during the trial. Billing starts only when the trial ends; cancel any time before with one click on your account page and pay nothing. If you forget, the plan price is charged when the trial ends and the plan renews until you cancel. One trial per person. Lifetime plans have no trial.

On the website account page under Your subscriptions, one click. Cancelling during the trial ends it at once with nothing charged. Cancelling later stops renewal and the plan keeps working until the end of the period you paid for.

Write to us within 7 days of a charge and it is refunded in full (first payment after a trial, renewals you did not mean to make, 30-day and 1-year passes, and Lifetime plans). A duplicate charge, wrong amount, a charge after you cancelled, or a charge that did not give you the plan is always refunded whenever you tell us. Refunds go to the original payment method and show in 5 to 7 working days. A person handles refunds: create a support ticket in Support
, or use the website contact page, and include the email of your Mimicore account and the payment id from your receipt.

Payments are processed by Razorpay. In India: UPI, cards and netbanking. Cards issued outside India cannot be used yet, because international card payments are still being enabled for Mimicore: if you are outside India, create a support ticket in Support
(or use the website contact page) and we will help you get your plan. Prices show in INR for customers in India and in USD elsewhere.

Razorpay retries a failed payment for a few days. If it still fails the subscription is paused and you are not charged. Common causes: an expired card, a low balance, or your bank blocking the recurring charge. Update your payment method or subscribe again on your account page. If the message says International cards are not supported, your card was issued outside India: those cards cannot be used yet (see that question). For a charge that went through but did not give you the plan, create a support ticket in Support.

This message means your bank or card is from outside India. Payments with cards issued outside India are not enabled for Mimicore yet, so trying again will not work. If money was taken for a failed attempt, your bank returns it automatically, usually within 5 to 7 working days. Create a support ticket in Support
(or write to us from the website contact page) with the email of your Mimicore account: a person will reply by email and help you get your plan.

Open Settings > Account and press Refresh, or reload the website account page. A plan usually appears within a minute after payment. If it does not, create a support ticket in Support
with the email of your account and your payment id.

On the website account page under Your invoices: open an invoice to view or print it. A receipt is also emailed after each payment.

Sign in on the website account page, type the code in the box under Choose your plan and press Redeem a free code. A gift code gives Studio for a number of days or for life, and works once per account. Discount codes (percent off) are different: they apply when you pick a plan at checkout.

Check for typing mistakes (codes are not case sensitive). "You already redeemed this code" means this account used it. "Already been used up" means the code reached its number of uses. "Expired" means its redeem-by date passed. "This is a discount code" means use it at checkout when you choose a plan instead.

You can install and sign in to Mimicore on all your PCs. Paid plans allow 1 active PC at a time (concurrent use). If you start converting or streaming on a new PC, it takes priority and pauses active processing on the previous PC.

Free live AI minutes (60 a day) refill every day. Studio removes the limit. Effects and the soundboard never use free minutes.

There is one free trial per person and per PC. After the trial you can still subscribe, or use the Free plan. If you think this is a mistake, create a support ticket in Support.

No. Live voice, cloning, effects, speech, dubbing and on-PC training all run on your PC. Microphone audio is processed in memory and never stored or sent. Typed scripts also stay on your PC, except when you are signed in and use the optional AI script polish or random narration in Audiobook Studio: that script goes to our server and an AI provider to be written. We store only your account email, plan, payment records and simple usage counters (which features you use and for how long, never any content; in stable builds you can switch this off in Settings > Privacy & data > Share usage statistics; early builds for testers send automatic technical diagnostics; the regular download does not). Optional cloud training uses only your own Google Drive and Colab, never our servers. In stable builds error reports are off unless you turn them on; early builds for testers send automatic technical diagnostics, and a support ticket sends only what you approve.

Your email address, sign-in method, plan and subscription status, free minutes used today, payment records (payment id, plan, amount, time), for Studio the list of your PCs, and simple usage counters (which features you use, how often and for how long, the kind of voice, the Speak/Dub language, graphics card vendor and app version) that you can switch off in stable builds in Settings > Privacy & data > Share usage statistics. Early builds for testers (beta versions, not the regular download) also send automatic technical diagnostics, retained for seven days, without a sharing switch. The counters are kept up to 400 days, are in your data export and are deleted with your account. We never receive your audio, voices, recordings or the text you type. Details: the privacy policy on the website.

In Mimicore: Settings > Privacy & data > Export my account data, or Delete my account. The same is on the website account page. Cancel any active subscription first. Deleting removes your plan, sign-in, tickets, chats and logs; payment records stay for accounting without a link to your account. Voices and files on your PC are not affected.

Settings > Privacy & data > Send error reports (off by default). Separately, anonymous performance diagnostics (technical numbers only) are on by default in every Windows build and can be switched off in Settings > Privacy & data. When on, a crash sends a short technical report (error message, app version, Windows version, a random installation id). Never audio, voices, text, your email or your account. Kept for 90 days. It helps us fix bugs.

Settings > Privacy & data > Copy support info copies versions, graphics card, key settings and the last log lines, with your Windows user name and email addresses removed. Paste it into an email. In Support you can instead create a support ticket in Support and approve the log that gets attached.

Only your own voice, or a voice whose owner gave you permission. Mimicore marks generated files as AI-generated. Impersonation, fraud and deception are forbidden by the Acceptable Use Policy.

Mimicore is for people 18 or older.

press Create a support ticket on the Support page and I will open a ticket for you; you get an email at every step. The website contact page lists the support email for billing, refunds and security reports.

Press Send feedback on the Support page or in Settings, About, or use the feedback page on the website (no account needed). Choose Idea, Problem, Praise, Question or Other, add a 1 to 5 rating if you like, and write your message. An e-mail address is optional and only used to answer you if you tick the box that allows it; Attach technical details is optional and sends only your app and Windows version, graphics card type and Live speed numbers, never audio, file names or logs. If you are offline the app keeps the message and sends it later. Feedback is read by the team and kept for 12 months; it is not a support ticket, so for a problem that needs solving use Create a support ticket.

On the Support page, describe what you did and roughly when, then press Create a support ticket. Mimicore picks the matching part of its local log, removes your Windows user name and email addresses, and shows you what it would attach; nothing is sent until you press Send ticket with log (or Send without log). You can follow the ticket in Your tickets and get emails at every step.

You get an email right away. Our automatic engineers analyse it: if it is not a defect you receive an explanation by email; if it is a real defect it is confirmed to you, a fix is built and tested, and you get an email with an early build to test. Press It is fixed or Still broken on your ticket. A person steps in if the automatic fix cannot solve it.

It can take some time because every fix is tested before it reaches anyone. You get an email at each step, and your ticket's status is always shown in Support under Your tickets. I cannot promise a date.

Restart Mimicore. If it keeps failing, open Support and press Create a support ticket (the log from the crash is attached with your approval), or copy support info from Settings if the app opens. If a new version fails to start three times, Mimicore rolls back to the previous release by itself and does not offer that version again.

Mimicore notes when it is running and removes the note when it closes normally. If the note is still there at the next start, it shows this message once. Your voices, projects and settings are not touched. Choose Open normally, Start in safe mode, or Send feedback with details (a pre-filled problem report with a short crash code and your technical details switch on; nothing is sent until you press Send). After two such closes in a row Mimicore starts in safe mode by itself so that a start problem cannot repeat. Safe mode switches off Live, the graphics card, hotkeys, the tray extras, update checks and diagnostics uploads; your voices list, voice import and export, Settings, Help and Send feedback still work. A banner says "Safe mode: Live is switched off. Restart to leave it." and has a Restart now button. You can also start safe mode yourself: hold Shift while Mimicore starts, or start it with --safe. Shutting Windows down or a power cut is not reported as a crash.

Mimicore needs you to be signed in (it is free). Open Settings > Account > Sign in.

On the Free plan your own voices can be used live for 60 minutes a day; they refill tomorrow. Built-in and gallery voices have no limit. Studio removes the limit.

That feature (creating more than one instant voice, mixing, importing, training, or a clip longer than the Free limit: 3 minutes for Speak, 2 minutes for Dub and converting files) needs the Studio plan. See Settings > Account; you can start a 7-day free trial there.

Install the add-on in Settings > Add-ons: Video dubbing for Dub a video, Trained voices (RVC v2) for trained or imported voices.

Speak needs somewhere to play: turn on My speakers, or set up the virtual mic in Live > Output and turn on the Virtual microphone output.

The recording had too little speech. Read the sentence aloud (about 5 to 12 seconds for calibration, 15 to 20 seconds for a voice sample) closer to the microphone and try again.

Quiet: speak closer or raise the input level in Windows sound settings. Clipping: move back a little or lower the level.

Some heavy things cannot share one PC: the live voice and Speak/Dub/conversions take turns (stop the live voice first), and one local training runs at a time. Training is the exception you can stack: one on This PC plus one in the cloud run in parallel.

No. Links were removed. Use Voices > the voice's... menu > "Export voice..." to save one.mimivoice file, send that file to the person (it contains the voice, so only send it to people allowed to use it), and they use "Import voice..." on the Voices page. By default the reference recording is included for your own voices and left out for voices of other people; the switch in the export dialog changes it. If someone gives you a web link, ask them for the.mimivoice file instead.

Yes: press "Train in the background" - training continues (on This PC and/or in the cloud) while you use Live, Speak and Dub; the Train dialog shows progress any time. One local job plus one cloud job can run together.

Settings > About (Updates) > "Check for early release". Early builds fix reported bugs and are only installed when you ask; the app never installs them by itself.

Free up space on the drive where Mimicore keeps its data (%LOCALAPPDATA%) - Settings > Storage shows what uses the space and cleans up what can be made again - then start the training again.

Mix voices works with instant voices only. Trained or imported RVC voices cannot be mixed yet.

That file is too large for this field. Use a smaller recording or text file.

On the website account page (sign in with the same account as the app): plan and trial, subscriptions with cancel, invoices, your Studio PCs, redeem a code, export or delete your account.

Yes: the Cloud training page on the website lets you start free cloud training in your own Google Drive and Colab without opening the app first.

On the website under Legal: License (EULA), Privacy, Acceptable Use and Refund policy, linked from the footer and the contact page.

Mimicore works with any PC app or game that lets you choose a microphone: set CABLE Output (VB-Audio Virtual Cable) as the microphone or input device in that app, or set it as the Windows default recording device. It does not touch the game itself. It does not work for phone calls on a phone.

Mimicore does not modify or inject into any game: it only provides a virtual microphone. Whether a game or server allows voice changers is up to that game's rules, so check them. Never use a voice to deceive, harass or impersonate anyone: see the Acceptable Use Policy.

Use a gallery voice (19 original voices, free), an effect such as Chipmunk or Giant, or create an instant or trained voice of a person who gave you permission. Trained voices also have a Pitch control (semitones). Mimicore does not change a voice's gender by a single switch.

On the Free plan, live voice changing with the built-in and gallery voices has no time limit, and your own voices can be used live for 60 minutes per day (it refills daily); effects and the soundboard have no limit. Studio has no live limit.

Mimicore changes the voice on your PC, so it works in calling apps on your PC (Zoom, Teams, Discord, Skype, WhatsApp Desktop) with CABLE Output as the microphone. It does not work on your phone or for normal phone-network calls.

Memory use varies with the installed model, voice and task. Effects need little processing. If Live says Struggling, check processing time and sound gaps. For trained voices, stop and compare positions on the latency and quality slider, or try a lighter voice, wired audio or a free effect. Your choice never changes automatically.

Yes: choose it in Live > Microphone. Bluetooth headsets add their own delay, so a wired headset is better for live use.

Settings > General > Appearance & language: choose dark, light or match Windows, and the language (English, Hindi, Spanish, Portuguese, French, German, Russian, Japanese or Indonesian).

Yes: Settings > General > Window: turn on Closing the window keeps the app running in the tray. Quit from the tray icon. Hotkeys and the live voice keep working while it runs.

If closing the window only hides it in the tray, right-click the Mimicore tray icon and choose Quit.

Yes. Speak turns typed text into speech in your voice and saves it as a WAV; Dub a video translates a video's speech into another language in your voice. Use the Ultra tier for the most polished speech. Only use voices you have the rights to.

Your own voice and voices you have permission for can be used in your content. The License requires that you never present output as a real person's genuine speech when they did not say it, that you disclose AI-generated audio wherever the law or the context requires it, and that you do not remove the AI-generated marking Mimicore adds to saved files. Read the License on the website for the exact terms.

I do not know of one. Plans and prices are on the website account page. Create a support ticket in Support if you have a special request.

Not yet. Today you use Mimicore through its hotkeys and the CABLE Output virtual microphone in OBS or any other app.

The app's source code is private. The cloud-training notebook is open source at github.com/ribanixlabs/mimicore-app.

Mimicore is made by Ribanix Labs. The tagline is Clone. Change. Create.

Fast works with 2 GB or more of graphics memory, or on the CPU: the lowest lag, for gaming and low-spec PCs. Balanced is best with 4 GB or more of graphics memory and is recommended for most PCs. Ultra (studio-grade 10-step decoder) wants 6 GB or more and is about 3 times slower; the live voice always uses the Balanced decoder. Any DirectX 12 graphics card from NVIDIA, AMD or Intel works through DirectML.

VB-CABLE is a free virtual audio cable by VB-Audio (vb-audio.com), donationware you can support on their site. Mimicore downloads it from vb-audio.com, checks that the installer is genuinely signed by VB-Audio, and installs it after Windows asks your permission. It creates the CABLE Output microphone that Discord, Zoom, OBS and games can use.

Keep the Colab tab open. If Colab asks whether you are still there, click it so the session continues. If there is no update for a while the Colab session may have stopped: open Colab and press the play button again; training continues from the last saved checkpoint.

When training finishes Mimicore installs the voice automatically, but it can take a while to download. You can start the install yourself from the cloud job in Voices > Train a voice > Cloud (press Install).

Mimicore downloads the voice models (Chatterbox by Resemble AI and RVC v2, MIT licensed) from Hugging Face and GitHub and checks every file against known checksums. They are kept on your computer and run with Microsoft ONNX Runtime and DirectML. Nothing you record is ever uploaded.

There is one free trial per person and per PC, so another account on this PC already used it. You can still use the Free plan, or subscribe to Studio on your account page. If you think this is a mistake, create a support ticket in Support.

Calibrate to my voice (trained voices only) shows a sentence on screen: read it in your normal speaking voice, about 5 to 12 seconds. Mimicore measures your pitch and your room, then sets pitch and noise gate for the closest match.

Speech synthesis and voice conversion run on your graphics card (the one picked in Settings > General > Processing), and text understanding runs on the CPU. The 28 effects run on the CPU.

Train a voice (Voices & effects > Train a voice) from 10 to 30 minutes of clean recordings of one person, in a quiet room, with the Good or Best quality. An instant voice from 15 to 20 seconds is quicker but less exact. Then use Calibrate to my voice on the Live page.

Windows privacy settings stop desktop apps from using the microphone. Open Windows Settings > Privacy and security > Microphone, turn on "Let desktop apps access your microphone" (and the main "Microphone access" switch), then press Start again in Mimicore. This is a Windows setting; Mimicore cannot change it for you.

The device you picked in Live is not plugged in or has been renamed. Pick another device in the Microphone or Output list (Windows default always works), press Refresh, and start again.

Mimicore needs somewhere to send the changed voice: switch on Hear myself (choose your speakers or headphones) or pick the virtual microphone (CABLE Input) under Output.

The audio devices did not answer in time. Close other apps that hold the microphone or speakers (a call, a recorder), choose the Windows default devices, and press Start again. If it happens again, create a support ticket in Support.

Free up space on the drive where Mimicore keeps the add-ons (C: by default; Settings > Storage can move them to another drive and clean up), or use Cloud (free) training instead, then install the Voice Training add-on again.

Open Settings > Add-ons > Cloud training and press Disconnect, then connect Google Drive again and tick the Google Drive box when Google asks. Mimicore only needs the permission to see the files it creates.

The cloud training is finishing. Wait a minute and try again; Mimicore installs the voice automatically when it appears.

Record longer, clearer speech: 15 to 20 seconds for an instant voice. Quiet rooms work best.

Dubbing translates sentence by sentence and idioms can come out literal, and short lines can be merged. Use clear speech in the original video, short sentences, and check the SRT subtitle file before you publish.

The gallery voices are original blends of English-speaking voices, so most carry English accents. For other accents create your own instant voice or train one.

Yes: choose monthly, annual or lifetime Studio under Choose your plan on the website account page. A plan change starts when your current plan's paid period ends, so you do not lose paid time. If something looks wrong, create a support ticket in Support.

Yes: to start a trial Razorpay charges a small refundable amount (shown at checkout in your currency) to verify your payment method and set up autopay. In India UPI autopay and cards work. Outside India, trials and autopay are not available yet. Cancel before the trial ends and you pay nothing.

Settings > Account > Sign out, then sign in with the other account (Google, Discord or email code).

Your account is tied to the email you signed in with (Google, Discord or an email code). To move to another address, sign in with that address (it is a separate account) or create a support ticket in Support and we will help.

Invoices are on your account page under Your invoices. The seller details on them are those of Ribanix Labs. Create a support ticket in Support if you need a specific invoice change.

Mimicore uses one microphone at a time: choose it in Live > Microphone. To bring in music or other sounds, use the Soundboard (clips play into the virtual microphone) or your streaming software's own mixer.

Always: clone only your own voice or the voice of a person who gave clear, informed permission for this specific purpose (written permission is recommended); get the speaker's consent yourself; tell people when speech is AI-generated whenever they could be misled; keep the AI-generated marking in saved files. Never: impersonate a real person without permission (including public figures, celebrities, family, colleagues, officials or support agents); commit or help with fraud, scams, phishing or extortion, or get past voice authentication (banks, phone systems); put words in a real person's mouth and present them as genuine; make robocalls or mass calls; create election, political or news deepfakes; create sexual or intimate content with a real person's voice without consent, or any content involving minors; clone a child's voice; harass, threaten or defame anyone; infringe copyright, personality, publicity or privacy rights; clone a voice taken from calls, videos or broadcasts of someone who has not agreed; or break any law that applies to you. Breaking these rules ends your right to use Mimicore.

Write to the support address on the website's contact page with the subject "Abuse report", or create a support ticket in Support
. Security problems go the same way; please give us a chance to fix them before sharing them publicly.

You can use Mimicore on all devices you own, with 1 active PC at a time. Account sharing with other people is not allowed. Creating several accounts to collect extra free trials is also not allowed.

The license is personal and non-transferable, for lawful use on devices you own or control. You may not sell, rent, lease or sublicense Mimicore. Invite and discount codes are personal and have no cash value.

Mimicore is provided as is, without warranty: voice cloning and conversion can be imperfect, delayed or fail. For quality problems try the Ultra tier, a cleaner recording, or training a voice; if something looks broken create a support ticket in Support.

Yes: your messages go to our server and, to write the answer, to a third-party AI provider (currently Groq). Please do not type passwords, payment details or other secrets. Chat messages are not stored except in a ticket (when one is created) and as a short rating record when you press Yes or No under an answer. A ticket contains the conversation and, only if you approve it on screen, a scrubbed part of the app log; it is analysed by an AI model from a third-party provider (currently Anthropic) to find and fix the problem. Tickets and ratings are deleted when you delete your account. See the Privacy policy on the website.

Prices show in INR for customers in India and in USD elsewhere: you only see the prices for your own country on the website account page. Where USD autopay is not yet available, USD customers buy one-time 30-day or 1-year passes that do not renew by themselves, and lifetime plans are one-time payments; the free trial applies to monthly and annual autopay plans. Payments are processed by Razorpay. Cards issued outside India cannot be used yet: create a support ticket in Support (or use the website contact page) and we will help you get your plan.

A subscription renews by itself every month or year until you cancel. A one-time pass (offered where autopay is not available) gives 30 days or 1 year once and does not renew. Lifetime is a one-time payment. You can see which one you have under Your subscriptions on the website account page.

An instant voice is made from about 15 to 20 seconds of your own speech and is ready in under a minute: it works in Speak, Dub and file conversion, and in Live too with the Instant live voices add-on (about 1 second behind you; needs a DirectX 12 graphics card). A trained voice is built from 10 to 30 minutes of clean recordings of one person (on your NVIDIA card or free in the cloud), gives the closest likeness and works live as well as offline; you can also calibrate its pitch to match you. The Free plan includes one instant voice and 60 minutes a day of live use with your own voices; more than one instant voice, trained voices, mixing and importing need Studio.

No card is needed for the 28 effects and the soundboard (they run on the CPU). For the live voice, a DirectX 12 graphics card from NVIDIA, AMD or Intel is recommended (Fast tier: 2 GB or more of graphics memory, or the CPU). Training your own voice on the PC needs an NVIDIA card (GTX 16 or RTX 20 series or newer, 4 GB or more); without one, use the free cloud training.

On the Soundboard page turn on "Also play on my speakers" (and keep "Play into the virtual mic" on so other people hear them too). This is separate from "Hear myself" on the Live page, which plays your changed voice.

Settings > Account shows your plan with a trial tag and the date it runs until; on the website account page, Your subscriptions shows the trial end date and whether it will renew. You also get an email when the trial starts and another two days before it ends.

Prices are shown in INR for customers in India and in USD everywhere else; there are no euro or pound prices. Payments are processed by Razorpay; cards issued outside India cannot be used yet (see Payment failed: International cards are not supported).

Mimicore has no password of its own: you sign in with Google, Discord or a one-time code sent to your email. To change a password, change it with Google or Discord. If the sign-in code does not arrive, check spam and ask for a new code.

When you create a support ticket from the Support tab, the card shows exactly what will be sent. The app log around the problem and basic system details (app version, Windows version, graphics card; your Windows user name and email addresses are removed, and there is never any audio in it) are included automatically, and you can untick that. Mimibot may ask for a screenshot (for something you can see on screen) or a short recording (for how a voice sounds): press Paste a screenshot after Win+Shift+S, Choose image…, or on the Live page press Record last 15 s and use Attach my latest live clip. You can also add a file yourself with Add a file…. Only images, audio and text files up to 8 MB, three files at most. Attachments are private, are opened only by support staff, and are deleted 45 days after the ticket was last updated.

In Settings > Privacy & data > Testing log, press Start testing log before testing. Test Live at a few slider positions, Speak, Dub, voices, soundboard and settings. Press Mark issue immediately after a problem. Logging resumes after an app restart. At the end press Stop testing log, then Export testing log and choose a ZIP. Review the ZIP before sharing it. It contains app events, command names/durations, system details, errors and Live timing, with no audio attachments or typed scripts; credentials and account identifiers are redacted, but device/file names can remain. The ZIP is not uploaded automatically. Up to 128 MB per session and four retained files in %LOCALAPPDATA%\Mimicore\testing-logs (or MIMICORE_HOME). Export before moving older session files elsewhere. Detailed logging can add overhead.

Yes, performance diagnostics are on by default in the Windows app (since 2.9.1 stable and early builds send the same) and you can switch them off at any time in Settings > Privacy & data. They contain technical data only: processing times, error counts and short redacted error messages, the names of the features that ran, app and Windows versions, the graphics card name, the power source, battery level and Battery saver state, and the random id of your Mimicore installation. While you are signed in they are linked to your account, so support can look into a problem you report, and they are exported or deleted with your account; a PC that is not signed in sends only the installation id. Audio, recordings, scripts, credentials, file and voice names are never included. Detailed records are kept 3 days and summed-up statistics 60 days, within per-PC and daily limits, so records can be dropped. Version 2.9.0 sent stable reports without the account, with only the graphics card family and a monthly PC id. Full local testing ZIPs are separate and are not automatically uploaded. A bounded outbox retries after offline sessions and restarts.

Yes. When an update has been downloaded, the banner and Settings > About > Updates offer Restart and install, and Install when I quit. With Install when I quit, the update installs after you close Mimicore (Quit from the tray menu, or closing the window when "keeps running in the tray" is off) and Mimicore does not reopen by itself. It never installs while Live, a job or a training is running or was running when you quit; in that case nothing is installed and the update is offered again next time. You can turn it off again with Don't install when I quit. Microsoft Store builds are updated by the Store and do not show it.

No. Press Download again (also after restarting Mimicore or the PC): the part that is already on your PC is kept and the download continues where it stopped. At the end every file is checked against its known checksum; only a file that fails the check is deleted and fetched again. After a failed attempt Mimicore retries by itself with growing pauses, and the progress shows the speed and the time left.

Yes. Download the add-on files on another PC or earlier (the same files Mimicore would download), copy them to this PC, then open Settings > Add-ons and press Install from a folder... (a folder that holds the files) or Install from files... (the files themselves) on that add-on. Mimicore checks the size and SHA-256 checksum of every file while copying it, refuses a file that does not match (it may be damaged or from another version) and tells you which files are still missing so you can add them or download the rest. Network locations are not accepted: copy the files to this PC first.

Mimicore uses the manual proxy from Windows settings (Network > Proxy) or the HTTPS_PROXY environment variable. If a proxy refuses or asks for a sign-in, the download stops at once with "A proxy blocked the download". Sign in to the proxy, ask the network owner to allow the download address (Hugging Face and GitHub), or install from files you downloaded elsewhere. Automatic proxy scripts (PAC) are not read.

Yes. The Mimicore button on the Windows taskbar shows the progress of downloads, voice training and renders, and flashes when one finishes while Mimicore is in the background. For a voice training or a render that took 20 seconds or more, Windows also shows a notification when Mimicore is not in front. Turn the notification off in Settings > General > Notify me when long jobs finish (on by default).