Transcription languages (speech → text)
When you upload a recording, the language is detected automatically — you do not usually need to set it. A wide range of languages is supported for turning speech into text in its original language.
Translation languages (text → text)
After a transcript exists, you can translate it into any of 30+ target languages. Translation is a separate step from transcription, so errors in the source transcript carry into the translation — review the source first.
Translation is available into: English, Spanish, French, German, Italian, Portuguese, Dutch, Russian, Ukrainian, Polish, Czech, Romanian, Hungarian, Bulgarian, Greek, Swedish, Norwegian, Danish, Finnish, Turkish, Arabic, Hebrew, Hindi, Chinese, Japanese, Korean, Thai, Vietnamese, Indonesian, Malay, and more.
Transcription vs translation
| Transcription | Translation | |
|---|---|---|
| What it does | Speech → text in the same language | Text → text in another language |
| When it runs | On upload (auto language detection) | After a transcript exists |
| Main risk | Mishearing names, numbers, jargon | Carrying source errors into the target language |
| Best practice | Review against the audio | Review after checking the source transcript |
Frequently Asked Questions
Do I have to set the language before transcribing?
Usually not — the spoken language is detected automatically. You can review the result and correct anything the model misheard.
How many languages can I translate into?
Transcripts can be translated into 30+ languages, including the major European, Asian, and Middle-Eastern languages.
Can it transcribe audio that mixes two languages?
Mixed-language audio is harder for any model and may need manual review at the points where the language switches.
Is the transcript language the same as the translation language?
Not necessarily. Transcription keeps the original spoken language; translation converts that transcript into a target language you choose.
Which languages are most accurate?
Widely-spoken languages recorded on clean audio transcribe most accurately. Accents, dialects, and noisy recordings reduce accuracy in any language.
Related resources
Reviewed by the TranscribeThis product team · Last updated: July 2026