Help

Accent in the translation: where it comes from and how to remove it

Updated 12 November 2026

In short

A voice clone learns from a person's speech in their native language. When that voice speaks the target language, it carries the pronunciation of the recording, and you may hear an accent. This is a property of cloning, not a bug. For many voices the accent is barely noticeable.

How to choose the pronunciation #

In the editor, the "Speakers" tab, every clone has two options (example for English to Russian):

  • "Speaker-1 · accent EN" — as in the recording. The voice sounds as close to the original as possible. This is the default.
  • "Speaker-1 · clean RU" — like a native speaker of the target language, but the voice sounds less alike.

Each option has ▶: listen to a phrase from your translation and choose. The speaker's phrases are voiced again, as when you change a voice. The clean option is available on Pro and Business. Your own voices from "Voices" work the same way.

Clean pronunciation noticeably lowers similarity. If the voice sounds fine with the accent option, keep it.

Where there is no clean option #

Clean pronunciation works when both the recording language and the target language are on this list: English, Russian, German, French, Spanish, Italian, Portuguese, Polish, Dutch, Hindi, Arabic, Turkish, Chinese, Japanese, Korean, Tamil, Czech, Swedish, Slovak, Hebrew, Thai, Telugu. For other languages the "clean" option is grey and marked "unavailable".

If the accent is in the way and you do not need similarity, pick a voice from the "Male" or "Female" library: these are native speakers of the target language.

See also Why the voice does not sound like the original.

Was this helpful?
Didn't find an answer?
Write in Telegram with your account ID, or email support@speeek.io
Contact support