How to Clone a Voice

Voice cloning lets you create a custom AI voice from a short audio recording. Once created, you can use it exactly like any other voice on VocalLab.

Watch the full walkthrough

Prefer to read? The written steps are below.

Before you start

• Cloning is free to try. A free VocalLab account includes 1 clone slot and a one-time 60 free points (about a minute of audio) — enough to clone a voice and hear it speak before paying anything
• You need a short audio recording of the voice you want to clone — 15 to 30 seconds (30 seconds recommended)
• The voice must be your own, or you must have explicit permission from the voice owner

Step 1 — Name your clone & pick a language

1. In VocalLab Studio, click Create a voice in the sidebar and choose Voice Cloning (or go straight to app.vocallab.ai/voice-cloning)
2. Enter a Name for your clone (e.g. "My Narrator")
3. Select the Language the voice will speak — this affects the accent and phoneme model used

Over 470 languages are available, and the list is searchable — type a few letters instead of scrolling. It starts with the fifteen that have a full voice catalog behind them (English, Spanish, French, German, Dutch, Portuguese, Italian, Japanese, Korean, Chinese, Russian, Arabic, Hindi, Hebrew, Polish), then everything else alphabetically.

Accent

Some languages show an Accent select next to Language. This is the difference between a Brazilian and a European Portuguese clone, or a British and an Australian English one — the same distinction that used to be impossible to ask for.

English offers 21 accents, Arabic 20, Spanish 9, Chinese 4, French 3, and Portuguese, Dutch and Bengali 2 each. Every other language has a single variant, so no Accent select appears.

Leave Accent on its default and the model decides from your sample.

Not sure of the language?

Pick Auto-detect from sample at the top of the list and the engine works it out from your recording. That was the old Other option — it still exists, but it's no longer the only way to reach a language outside the core fifteen, since they're all directly selectable now.

Note: languages without a stock voice catalog are marked Experimental — cloning works, but quality varies more than it does in the fifteen main languages.

Step 2 — Add your audio sample

Under Sample source, pick one of two options:

Record

• Click Start recording and allow microphone access
• Read the suggested text aloud naturally — recording stops automatically at 30 seconds
• Click Stop recording when done

Upload

• Drag and drop or browse for one file
• Accepted formats: MP3, WAV, WebM
• Maximum file size: 4 MB
• Duration: 15–30 seconds, 30 s recommended (longer files are auto-trimmed to 30 s)

Transcript (optional)

A Transcript of exactly what's said helps the model clone the voice more accurately. Type it in, or click Auto-transcribe (this uses a few points, like Speech to Text).

What makes a good sample

• One clear voice, no background music or noise
• Normal speaking pace and volume — not whispering or shouting
• A quiet room with minimal echo
• Recorded close to the microphone

Optional settings

• Remove background noise cleans up the sample before cloning. Recommended if your recording isn't from a quiet room.
• More settings lets you add tags and a description to help you find the voice later.

Consent

Tick the consent box to confirm you have the right to clone this voice. This is required before proceeding.

Click Continue when ready.

Step 3 — Processing

VocalLab prepares your audio and creates the voice. This usually takes under a minute. Do not close the tab.

Step 4 — Done

Your clone is ready. You'll see a success screen with two options:

• Use your voice now — start generating with it
• My voices — view all your voices

Creating a clone doesn't cost any points — it only uses one of your plan's clone slots.

Using your clone

Your clone appears in Voices → My Voices under the Cloned tab. Select it in the voice picker on any generation page and use it like any other voice.

Using your clone in another language

Yes — a clone can read text in a language other than the one you cloned it in. Just pick it in the voice picker and type in the other language.

Expect an accent. Your clone keeps the accent of the voice you cloned. A Russian speaker cloned in Russian will sound like a Russian speaker when reading English; an English clone reading Spanish will sound like an English speaker reading Spanish. For many people that's exactly the point — it's a fast way to get an authentic-sounding accented voice.

If you want it to sound native rather than accented, use Regenerate in another language on a finished generation: it translates the text and has your clone speak it in the target language.

Tip: switch the model to VocalLab Studio for non-English output — it covers 200+ languages, where VocalLab Pro covers the core 15.

Clone limits

On paid plans you can create as many clones as you like — the limit is how many you can keep saved at the same time. Delete a saved clone to free a slot and clone again.

Free plan: the 1 free clone is a one-time allowance, not a slot you can reuse. Deleting it does not give it back, and buying extra points doesn't add clone slots either. To replace a clone, upgrade to any paid plan (Lite and up).

Lite users with 1 existing clone must delete it before creating a new one. Free users can't replace their clone by deleting it — see above.

Troubleshooting

"Add one audio sample to create your voice." Record or upload a sample before clicking Continue.

"At least 5 seconds of audio is required." Your recording is under 5 seconds. Record or upload something longer.

File not accepted Only MP3, WAV, and WebM files are supported. Convert other formats (M4A, AAC, FLAC) to MP3 or WAV first.

"Your plan includes a single cloned voice. Delete it to create a new one, or upgrade your plan for more." The Lite plan allows 1 clone at a time. Go to Voices → My Voices, delete your existing clone, then try again. On the Free plan, deleting won't help — your one free clone is used up once it's created, so upgrade to clone again.

Clone sounds robotic or wrong language Make sure you selected the correct language in Step 1. Re-create the clone with a cleaner, longer sample (30 seconds works best).

Microphone access denied Allow microphone access in your browser settings, then reload the page.