OCR reads the source
The OCR provider estimates the visible characters and their order from the image.
Supported Image Translation Languages
See the target languages available in the image translator, how OCR source recognition differs from translation, and where mixed scripts need extra review.
The short answer
The translator offers 38 target-language choices. Source-text recognition is a separate OCR step, so a listed target does not guarantee that every photo, script, or layout will be read correctly.
Three language decisions happen in one request
Context identifies the language
A full sentence gives stronger evidence than a name, code, or one-word button.
Translation writes the target
The selected target tells the translation stage which language to produce.
For a short or mixed-language source, review whether the photo translator can identify the source language before trusting the result.
When a short public notice needs location and symbol context, follow the workflow to translate a sign from a photo.
Available target languages
The groups below make the list easier to scan. They do not rank quality or claim that every language uses only one script.
Latin alphabet targets
Common European and Southeast Asian target languages that mainly use the Latin alphabet.
- English
- Spanish
- French
- German
- Italian
- Portuguese
- Dutch
- Polish
- Turkish
- Vietnamese
- Indonesian
- Malay
- Czech
- Danish
- Finnish
- Norwegian
- Swedish
- Romanian
- Hungarian
- Croatian
Cyrillic target languages
Languages offered here that commonly use Cyrillic, while some may also use another script.
- Russian
- Ukrainian
- Bulgarian
- Serbian
Right-to-left targets
These languages need right-to-left text handling even when numbers or Latin terms appear inside a line.
- Arabic
- Hebrew
- Persian
- Urdu
South Asian targets
These options use distinct scripts and need clear marks above, below, or around the main character shapes.
- Hindi
- Bengali
- Tamil
- Telugu
East Asian targets
Character context, writing direction, names, and mixed Latin labels need careful review.
- Simplified Chinese
- Traditional Chinese
- Japanese
- Korean
Other script targets
Greek and Thai each use a distinct script that must remain clear at the OCR stage.
- Greek
- Thai
Match the workflow to the writing system
| Situation | Main risk | Best next step |
|---|---|---|
| Arabic, Hebrew, Persian, or Urdu | Mixed right-to-left text, numbers, and Latin names can change reading order. | Use the guide to translate Arabic and Hebrew images. |
| Chinese, Japanese, or Korean | Names, readings, and vertical regions may need separate treatment. | Follow the CJK image translation workflow. |
| Names, addresses, dishes, or brands | A useful identifier may be translated when it should keep its sound or official spelling. | Compare transliteration with translation. |
| Several scripts in one image | OCR may join regions in the wrong order or assign one language to the whole page. | Separate regions with the mixed-script workflow. |
What the list does not promise
- A target-language option does not certify OCR support for every source script, font, or handwriting style.
- Names, codes, measurements, and official spellings still need direct comparison with the image.
- Legal, medical, identity, financial, and safety text still needs the level of human review required by its use.
When the source is clear and the target is selected, open the AI image translator and keep the extracted text beside the translated result.