In addition to the regular API, we added 3 special endpoints, specifically designed to be really fast while doing a lot of work:

Using these endpoints and our complete guide, you are now able to create a library for your own stack, as we did for the existing libraries.

If you created your own library, please contact us so we can add it on our listing. We can also help you maintain it and host it on https://github.com/translation for more visibility.

We created a new API that uses REST principles and conventions. It receives JSON parameters and returns JSON responses.

The main page is located here: https://translation.io/docs/api

Using the following models, you can very easily create custom translation workflows:

• Segments
• Tags
• Source Editions

We intend to keep on adding more models in the future, so you will be able to manage all of your resources programmatically.

If you have any feedback regarding our API, please contact us at contact@translation.io

Sometimes you have no choice but to confront your translators with HTML or interpolated variables.

The problem is that translators do not necessarily know the meaning of these notations and may be tempted to translate them or may inadvertently alter them, as shown here below:

In the first example, the name variable will not be replaced by the username in French. In the second, the HTML will not be correctly interpreted and the page will be broken.

To help translators manage these notations more easily, we now highlight HTML tags and interpolated variables. We also display a warning icon if some of these tags and variables are missing or incorrectly translated.

On top of that, translators are now able to insert these notations into the target field by clicking on them or pressing the TAB key. This really speeds up the translation process and prevents unfortunate typos.

Machine Translation (MT) is a term used to describe translations that are carried out by a computer. One very popular MT service is Google Translate.

Recently, Google Translate started to use Deep Learning strategies to improve its translations with great success. Even though Google Translate is still not comparable to a native-speaker translator, the results are usually “good enough” until someone proofread them.

For this reason, we created the new Auto-Translate feature⁽¹⁾, which you will find in the project settings.

If you enable it, every new source text will be automatically translated for free using Google Translate, and then pushed back into your code. A new auto translated tag will be added to these segments and your translators will be notified that they may proofread them at their own convenience.

You can also click on “Auto-Translate now” to translate an existing project into a specific target language.

You may want to add a custom language that is derived from an existing language. It's useful if you want to adapt some translations to another instance of your application, or to a specific customer.

The structure of a custom language is: existing language code + - + custom text, where custom text can only contain alphanumeric characters and -.

Examples: en-microsoft or fr-BE-custom.

Custom languages can be added like any other language.

Fallbacks work as expected. It means that if the en-microsoft.some_key is missing, then it will fallback to en.some_key. So you only need to translate keys that should be customized.

Note that fallback are chained, so fr-BE-custom will fallback to fr-BE that will itself fallback to fr.

Using GetText syntax, it will only fallback to the source language. So either you create a fallback mechanism by yourself or you avoid fallbacking by translating everything in Translation.io for that custom language.