How to clone your voice: record or upload a sample
In Settings → My Voices, tap Add new voice, read the text on screen or pick a file of 10 s to 5 min, then save. MiniMax costs 180 tokens, Qwen 3 about 1.
Updated November 2, 20263 min read
A saved voice lets you voice any text or a translation in your own voice. You need one clean recording of the person whose voice it is. Clone only your own voice or the voice of someone who has agreed to it.
How to open the New voice screen
Open the Settings tab.
Tap My Voices.
Tap Add new voice.
The same screen opens from the voice picker in Text to Speech (My voices tab → Add your own) and from the Translate audio settings (Add MiniMax voice or Add Qwen voice). The first time, the app may ask for permission to send recordings to AI services. Without it, voice cloning doesn't work.
How to record your voice
Under Provider, choose MiniMax or Qwen 3. How they differ: what a custom voice is and where it works.
Open the Record your voice tab.
Tap the microphone button. The first time, allow microphone access.
Read the text on the screen aloud at your normal pace. The timer shows how long the recording is. Reading the whole text takes about half a minute.
Tap the stop button.
Listen to the recording in the player. If you stumbled or something made noise, tap Record again.
Tap the save icon in the top right corner.
Enter a name for the voice, or keep the suggested one (My voice 1, My voice 2 and so on), and confirm.
How to use a file you already have
Any audio file with one person speaking will do, for example a voice message saved to your phone. The file must meet these limits:
| Parameter | Limit |
|---|---|
| Length | from 10 seconds to 5 minutes, 60 seconds recommended |
| Size | up to 20 MB |
| Format | audio only: MP3, WAV, AAC, OGG, M4A and other common formats |
Choose the provider, as in the first step above.
Open the Use existing file tab and tap Choose file.
Pick the file and listen to it in the player. To replace it, tap Choose another file.
Tap the save icon in the top right corner, enter a name and confirm.
What happens after you save
The screen closes and the voice appears in My Voices with the Cloning... badge. Cloning usually takes less than a minute, and you can leave the screen meanwhile. When the badge disappears, the voice is ready in Text to Speech and Translate audio.
MiniMax: 180 tokens are charged when cloning starts and come back if it fails. Your balance needs at least 180 tokens.
Qwen 3: about 1 token.
If the badge changes to Failed, see voice cloning failed or doesn't sound like you.
If the app won't save the recording
The app checks the recording when you tap save. These messages mean the recording didn't pass the check, and no tokens were charged.
| Message | What to do |
|---|---|
| Record your voice first | Record your voice or choose a file before saving. |
| Recording is too short. Minimum duration is 10 seconds. | Record again and read the whole text. |
| Recording is too long. Maximum duration is 5 minutes. | Trim the file or record a shorter part. A minute is enough. |
| File is too large. Maximum size is 20 MB. | Choose a shorter or compressed file, for example MP3 instead of WAV. |
| Could not access microphone | Allow microphone access: see microphone or camera access denied. |
| Failed to save voice. Please try again. | Check your internet connection and tap save again. |
How to make the clone sound closer to you: how to record a good sample.
Contact support
Describe the problem and we will reply by email. Add the file name or job ID if you have one.