Here is the straight answer up front: Capture2Text has no translation API you can configure. Its translation feature runs on a built-in Google Translate connection — there is no API key field, no endpoint setting, and no DeepL integration anywhere in the program. This page explains exactly what the “Translation API” behind Capture2Text really is, how the target language is chosen, and the honest way to bring DeepL into your workflow.
Capture2Text v4.6.3 is free, open source (GPLv3), portable, and Windows-only. If you have not installed it yet, grab it on the download page. For the full picture of the feature itself, see the translate feature guide and the step-by-step Google Translate setup walkthrough. Browse all walkthroughs in the guides hub.
What Capture2Text’s translation actually uses
Every time you capture text with translation enabled, this happens behind the scenes:
- You select a screen region with the capture hotkey (default Win+Q).
- Tesseract OCR converts the pixels into text entirely on your PC — no internet needed for this step.
- Capture2Text sends that text to Google Translate over the internet and gets the translated result back.
- The translation appears in the popup window, is appended to your clipboard, or both — depending on your settings.
That is the entire “API”: a hardcoded Google Translate connection. Google provides the translation; Capture2Text just passes text back and forth. This is why translation is the only Capture2Text feature that requires an internet connection — the official documentation states this explicitly, and everything else (capture, OCR, clipboard, hotkeys) works fully offline.
Where the target language is set
There is no single “translate to” master switch. Instead, each installed OCR language gets its own target language:
- Right-click the Capture2Text tray icon and open Settings…
- Go to the Translate tab.
- For each OCR language you have installed, pick the language you want it translated into from the dropdown.
So if you OCR Japanese manga with the Japanese OCR language active, you would set the Japanese row’s target to English; if you also OCR German documents, the German row might target English too — or a different language. Each row is independent, which is genuinely useful if you work in several source languages.
The same tab holds the two output switches:
- Show translation in popup window — displays the translation alongside the original OCR text after each capture.
- Append translation to clipboard — adds the translation to your clipboard after the OCR text, separated by a separator string you can customize.
Two honest limitations from the official documentation: some OCR languages have no translation support at all and simply will not appear in the list, and if your machine is offline, translation silently does nothing while OCR keeps working.
Why there is no DeepL or custom API option
Short version: the developer built translation on Google Translate’s free tier, and never added a settings panel for anything else. The Translate tab offers language dropdowns and output toggles — there is no field for an API key, no endpoint URL box, and no provider selector. We verified this against the official documentation: the Translation chapter describes only the Translate tab options above, and nothing about custom APIs exists anywhere in the settings.
This is a reasonable design choice for a free portable tool: Google Translate covers 130+ languages with no signup, no key, and no billing — exactly what a zero-config screen OCR utility needs. DeepL integration would require each user to bring their own API key (DeepL’s free API tier has monthly character limits; paid plans start around US$8–9/month), which does not fit a “download and press Win+Q” tool. The tradeoff is quality nuance: DeepL is widely regarded as stronger for certain language pairs (notably German and other European languages), while Google Translate wins on language breadth.
To be explicit, because this question comes up constantly: Capture2Text has no affiliation with, endorsement from, or commercial relationship with Google, DeepL, or any translation provider. This site is an independent resource and is not partnered with any of them.
The correct way to use DeepL with Capture2Text
You cannot make Capture2Text call DeepL automatically — but you can absolutely use DeepL on text Capture2Text produces. The manual workflow takes seconds:
- Do your capture with Append translation to clipboard enabled, or just copy the OCR text from the popup.
- Paste the text into DeepL’s translator in your browser.
- Read the DeepL version when you need its nuance; keep the built-in Google version for quick glances.
Which text should you paste — the OCR original or the Google translation? Paste the original OCR text, not the English translation. Translating already-translated text (Japanese → Google English → DeepL German) compounds errors. The OCR result is the cleanest source you have, so feed that to DeepL directly.
For a deeper comparison of when each engine earns its place, read our honest DeepL vs Google Translate for OCR text breakdown.
What the “API” cannot do
- No custom endpoint. You cannot point Capture2Text at your own translation server or a self-hosted model.
- No API key. There is nowhere to enter a Google Cloud Translation API key or a DeepL API key — billing-based quota upgrades are not possible inside the app.
- No offline translation. The translation step always needs the internet. The OCR step does not.
- No per-capture language switching. The target language follows the active OCR language’s setting; switching target languages means opening Settings.
- No batch translation of files. Translation happens capture-by-capture. For documents, see the translate scanned documents workflow.
FAQ
Does Capture2Text use the Google Translate API or the website?
It uses Google’s translation backend directly from the application — you do not need a browser, an account, or an API key. From your perspective it is simply “built-in Google Translate”: free, unlimited for normal use, and tied to Google’s service availability.
Can I add my own translation API key to Capture2Text?
No. The settings contain no API key field and no endpoint configuration. If you need key-based translation (for higher quotas or billing), do the translation step in a separate tool and use Capture2Text purely for OCR.
Why is there no DeepL option in Capture2Text?
The developer implemented translation with Google Translate only, and the program has not been updated since v4.6.3 (2022-03-19). The practical workaround is copying OCR text into DeepL manually, which takes a few seconds per capture.
Is Capture2Text’s translation really free?
Yes. There is no account, no key, and no payment inside Capture2Text. The cost is simply that translation requires an internet connection and goes through Google’s servers — do not capture passwords or sensitive text with translation enabled.
Can I translate offline?
Not with Capture2Text’s built-in feature. OCR itself works offline, but the translation step needs internet access. For offline translation you would need a separate on-device translation tool applied to the OCR text.