Machine Translation

Localazy offers you the possibility to easily translate your digital content with the help of machine translation (MT) engines within the web interface.

The Free tier provides Amazon Translate suggestions to support you by default, but sometimes, it’s not enough. When you want to use multiple machine translation engines at once, Localazy includes the Additional MT Engines feature.

Localazy helps you make better choices by displaying suggestions from multiple MT engines simultaneously, allowing you to make a more precise and accurate translation even when you don’t fully understand the target language.

Additional MT Engines feature is available with the Autopilot plan and higher tiers.

Supported machine translation engines are:

• Amazon Translate (default)
• Google Translate
• DeepL
• Azure Translator

We can gauge how good a Machine Translation Engine is by the accuracy of the translation, conciseness, consistency of output, and how many language pairs they offer.

Localazy offers four major machine translation engines to use when translating your content (available with the Additional MT feature enabled).

Use English, which is unquestionably the best language to pair with another language for machine translations because translations from and to English give the best results overall. That’s why it is also a good idea to select English as the source language for your projects in Localazy.

As for the engine itself, each one has its pros and cons:

• Google Translate is the most popular and offers many languages and language pairs with a reasonably accurate translation, making it the best average option.
• DeepL is becoming quickly better and better and is great for more technical terms. It is usually also very concise but has a lower variety of language pairs.
• Amazon Translate still has a somewhat fluctuating quality; it can be very high or pretty low, but you can use it for free with Localazy.
• Microsoft Azure has fewer available language options and average quality.

The best thing you can do is combine and compare multiple machine translation outputs and decide on the best results with backward translations and other checks.

Machine Translation Post Editing (MTPE) is the process of reviewing and editing the content translated by the machine translation engines giving it a final, more accurate result.

MTPE can include checking that all placeholders are correct and that the project-specific content is translated correctly and consistently.

When you start with Localazy, you can work with Amazon Translate right off, as this engine is available in the Free tier by default.

If you want to use multiple machine translation engines, you can get the Additional MT Engines feature with the Autopilot plan and higher tiers.

Enable Additional MT Engines to get suggestions from:

• Amazon Translate (default)
• Google Translate
• DeepL
• Azure Translator

You might also want to read: #️⃣

Intro to Localazy from translator’s perspective

Machine Translations (MT) are an incredible invention that makes our life easier. But sometimes, their accuracy is low, and translating large amounts of texts will inevitably produce outputs of varying quality.

Translating strings to a language you don’t know well with MT engines can be a pain because you might not fully understand what’s truly in there.

What can you do to ensure consistent quality levels of machine translation output? Localazy offers some solutions you might want to consider:

• If you are conducting the translations yourself, try to use Additional MT Engines to get more suggestions, allowing you to compare and decide.
• If you don’t want to spend your time with machine translation post-editing and checking, try the Human-Assisted Machine Translations service by Localazy and get machine translations checked for a fair price per word.

To automatically translate your content using machine translations, you need to set up an Automation.

Navigate to the Tools -> Automations section in your project, create new, and select the action “Use machine translations”.

Configure the trigger on events such as key uploads or updates, and specify the target languages for machine translation.

Read more about content translation on blog.

Currently, there are two built-in engines that take your glossary terms into account when translating:

1. DeepL integration
2. Google Translate integration

While these integrations provide helpful translation capabilities, please note that the output of these engines and their models cannot be guaranteed to be 100% reliable. We cannot take responsibility for the translations they provide. The accuracy depends on the language pair (English as source is recommended) and the provider performance.

Another way to translate your strings automatically with glossary terms respected is to configure the OpenAI integration for your account.

Remember: It is always recommended to follow machine and AI translation with a human review. If you don’t have the workforce available for reviewing your MT translations in-house, you can activate our Human-Assisted Machine Translation service.

The advantages of machine translation are:

• Multiple free translation engines as options, like Google, DeepL, etc
• A myriad of language pairs, from English-Portuguese to Russian-Zulu
• Instantaneous translations. You can translate in real-time.
• Machine learning mechanisms are improving, meaning that every day the services are getting better.

The disadvantages of machine translation:

• Language pairing accuracy without English can be very low, giving it some limitations
• Risk of incorrect translation.
• Translation Engines struggle with context while translating.
• Mistakes might create irreversible situations.

How to mitigate the disadvantages of MT?

Accuracy, context, and mistakes are solvable by using Human-assisted Machine Translation and Machine Translation + Post-Editing.

You can employ both of these methods while using Localazy to achieve the best possible quality of machine translations.

Machine translation (MT) is the process by which computer software translates text from one language to another, without human intervention.

To set up OpenAI integration with Localazy:

1. Obtain Your OpenAI API Key – Sign up at the OpenAI platform, create/select a project, generate a secret API key, and set up billing.
2. Connect with Localazy AI – Go to Console -> Localazy AI, select your organization, enable AI assistance, and paste your OpenAI API key. Choose your preferred OpenAI model and save the settings.
3. Configure AI Options – Customize AI features for your project, including translation suggestions and bulk pre-translation.

Learn more about the Localazy AI options in the docs

There are some limitations to consider when using Localazy¨s built-in machine translations:

1. Machine translation might not work in case your source content is too long. (Learn more)
2. Machine translation currently does not apply to plurals, because there might be different plural forms in the source and target language.
3. Machine translation does not translate keys with translations that are already translated, even though there might be another version waiting for review (waiting for review = true AND missing current version = false).
4. Machine translation might produce unexpected errors in your placeholders, glossary terms, and special markup contained in the source strings. Localazy prevents most of these issues automatically, but it is always advised to followup with a human review.
5. The Machine translation feature is also subject to a fair use limit of 2 million characters per month to ensure equitable access for all users. This limit is automatically applied to all accounts using the feature. If your translation needs exceed this threshold, Enterprise customers can access a higher limit tailored to their requirements. Please contact us to discuss Enterprise plans and expanded usage options.