Import limits
- +WAV, MP3, M4A, or FLAC up to 20 MB per file
- +Ten samples per account, ten seconds minimum each
- +One speaker per sample keeps the clone stable
Voice cloning studio
Clone a voice from ten seconds, then let it speak
Import a clean sample or record one in the browser. VoicesCloning mirrors the timbre, pace, and accent of the speaker, then renders your script in that voice across 32 languages - with noise removal and accent optimization on hand when the source recording is not perfect.
10-second reference32 output languagesNoise removal built inConsent-first voice slotsPreview before you render
Clone a voice in one pass
Import a clean voice clip or record 10-60 seconds in the browser, tune the cleanup options, write your script, and preview the cloned voice. Two voice slots are included with every account.
Engine voices-clone-v2
Import your voice
up to 20 MB per fileAdd or drop a fileWAV, MP3, M4A or FLAC · 10 seconds minimum
10-60 seconds long
Uploaded samples0/10
No samples yet. A clean recording with one speaker works far better than a noisy interview clip.
Text to preview
up to 300 charactersVoice slots used: 0/10English
Generated voice results appear here. Nothing is synthesized until you press Generate.
Recording tips
- +Record 10-60 seconds in a quiet room
- +Keep a steady distance from the microphone
- +Avoid reverb, music beds, and overlapping speakers
Rights and consent
- +Only clone voices you own or are licensed to use
- +Confirm the rights checkbox before every render
- +Delete a voice slot at any time to remove the embedding
Sample the range before you clone
Every voice below was produced with a ten-second reference clip. Preview a card to hear how a clone holds together across narration, advertising, and long-form reading.
Ava · Warm narratorEnglish (US) · mid-30s · documentary
10s reference
Kenji · Bright presenterJapanese · 20s · commercial
10s reference
Noor · Deep explainerArabic · 40s · corporate
10s reference
Elise · StorytellerFrench · 30s · audiobooks
10s reference
Marco · Casual hostSpanish (LatAm) · 20s · podcast
10s reference
Lin · Studio anchorChinese (Mandarin) · 30s · news
10s reference
Built for voices that have to hold up
Everything you need to clone, clean, steer, and ship a voice - without assembling a pipeline of separate tools.
Cloning from a short sample
Ten seconds of clean speech is enough to build a speaker embedding you can reuse across scripts, languages, and sessions.
32 languages, one identity
Write the script in any supported language and keep the same voice identity instead of hunting for a matching speaker.
Noise removal and accent control
Strip room tone, hiss, and hum from the reference, then decide whether the accent should follow the source or the target language.
Broadcast-ready output
Every preview is normalized to -16 LUFS with de-essing and loudness matching, so it drops straight into a timeline.
Voice design from a description
No reference recording? Describe age, gender, pace, and emotion to design a voice, then clone it like any other sample.
Script-to-caption workflow
Long scripts are split into natural sentences so pacing stays consistent across narration and dialogue.
Multi-speaker projects
Keep a different cloned voice per character and re-render a single line without touching the rest of the project.
Voice slots you control
Each slot stores one embedding. Two slots come with every account, and extras come with Pro and Studio.
Fast first render
Most previews finish in under 30 seconds, so you can compare takes before you commit to a full read.
From a raw recording to a finished read
Four steps, no timeline gymnastics. The reference audio is the only part that needs care.
Import or record
Add a clean sample or record 10-60 seconds in the browser, then let noise removal tidy the reference.
Write the script
Type up to 300 characters for a preview, or paste a longer script when you are ready for full narration.
Generate and compare
Render the clone, listen to the preview, and adjust pacing or accent settings until it sits right.
Ship the audio
Download the rendered file and reuse the same voice slot for every line in the project.
Match the input you already have
The studio handles a single cloned narrator as easily as a scripted multi-speaker scene.
| Workflow | What you provide | Best for |
|---|---|---|
| Cloned single voice | One clean 10-second sample | Narration, explainers, audiobooks |
| Designed voice | A written description instead of a recording | Campaigns that need a voice nobody owns |
| Multi-speaker scene | Several samples, one per character | Dialogue, drama, and scripted podcasts |
| Localized read | One sample plus a translated script | Ads and product videos for more than one market |
Where a cloned voice pays for itself
Creators, localization teams, and product teams use one voice slot across an entire catalogue.
Creators and podcasters
Keep a consistent host voice across episodes, intros, and ad reads without booking studio time.
episode-12.mp3 · host voice slotAudiobooks and narration
Cast a whole manuscript with cloned or designed voices and re-record a chapter line by line.
novel.epub · 24 chapters · 6 voicesLocalization teams
Keep the original speaker identity while the script moves into another language.
campaign-en.mp4 · ja + de + ptE-learning studios
Update a single slide of narration without re-recording the whole course.
module-03 · slide 14 re-renderAccessibility work
Turn written material into speech quickly, with a voice listeners already recognize.
newsletter.txt · 4,200 wordsGames and prototypes
Placeholder dialogue that still sounds like a real performance while the script is still moving.
scene-04 · 3 charactersStart free, scale when the audio ships
Two voice slots and preview renders are free. Pro and Studio add slots, longer renders, and commercial rights, with annual billing saving 17%.
Free
$0to start
Try the studio with two voice slots and preview renders.
Open the studio- 2 voice slots
- Up to 300 characters per render
- Noise removal and accent optimization
- Watermark-free previews
Pro
$12/month
For creators publishing cloned or designed voice work.
Choose Pro- 40 voice slots
- 10,000 characters per render
- 32 output languages
- Commercial rights for solo creators
- Batch rendering and re-render per line
Studio
$39/month
For teams shipping voice at volume.
Choose Studio- 200 voice slots
- 100,000 characters per render
- Unlimited languages and projects
- Team workspace with shared voice slots
- Extended commercial and client rights
Annual billing saves 17%. Plans are managed in your account area and can be cancelled at any time. Previews stay free; paid plans unlock longer renders, more voice slots, and commercial rights.
VoicesCloning FAQ
Short answers on sample length, languages, rights, and what each plan includes.
How long does a voice sample need to be?+
Ten seconds of clean, single-speaker speech is the minimum. Thirty to sixty seconds gives the embedding more prosody to work with, which usually sounds more natural on long reads.
Do I need to record in a studio?+
No, but the sample decides the result. A quiet room, a consistent distance from the microphone, and no music bed produce a far better clone than a noisy interview clip.
Can I clone a voice in another language?+
Yes. The voice identity stays with the speaker while the script can be written in any of the 32 supported output languages, with accent optimization on or off.
What happens to my samples?+
Samples are used to build the voice embedding for that slot and are stored with your account. Deleting a slot removes the embedding; our Privacy Policy covers retention in detail.
Do I need the speaker’s permission?+
Yes. You must own the voice or hold written permission from the speaker. The rights checkbox and our Terms of Service exist to keep that requirement explicit.
How many voices can I keep?+
Every account starts with two voice slots. Pro raises that to 40 and Studio to 200, and slots can be reassigned when you retire a voice.
Can I re-render just one line?+
Yes. Long projects keep the script, the voice slot, and the settings, so you can re-render a single sentence without regenerating the whole file.
Is there a watermark on previews?+
Previews render without an audible watermark. Usage rights are governed by your plan, not by an audio mark.
Which formats can I export?+
Previews and renders download as MP3 or WAV, with captions available as SRT or VTT when you work from a script.
How much text can one render cover?+
Free accounts render up to 300 characters per pass. Pro covers 10,000 characters and Studio covers 100,000, which is enough for a full chapter in one job.
Can I cancel whenever I want?+
Yes. Subscriptions are billed monthly or annually through Stripe and can be cancelled from your billing settings at any time.
Is this the official MiniMax website?+
No. voicescloning.lol is an independent third-party website and is not affiliated with, endorsed by, or operated by MiniMax or any other voice AI service.
Give your script a voice it can keep
Sign in with Google, pick a plan, and render your first cloned preview in the studio console.
Rights-first workflow 10-second minimum sample 32 languages Cancel anytime