What is MTPE?

It is a process in the translation industry where source text is translated using a machine translation engine, such as Google, DeepL, Microsoft, or Amazon, and then reviewed, edited, and approved by an experienced human translator.

Taking MTPE One Step Further

When machine translation engines like Google and DeepL were first introduced to the translation market, their capabilities were limited, and the quality of their output was often unsatisfactory. However, with advancements in translation technology, we can now create client-specific and industry-specific profiles using our translation management software. These profiles allow us to train machine translation engines to provide higher-quality output tailored to specific needs.

How Do We Train the Machine Translation Engine?

Once we have created a client or industry-specific profile, all related projects are assigned to that profile. We select the most suitable machine translation engines based on the target languages and enhance the engine’s accuracy by inputting human-translated content into the translation memory. Additionally, we integrate a glossary of terms that have been translated by human linguists and approved by the client. Other parameters, such as tone, industry sector, and style preferences, are also incorporated.

The more MTPE projects we complete under a specific profile, the more the AI learns how to translate effectively for that client or industry. After approximately 12 months of continuous training, we aim for an accuracy rate of 85-95%, depending on the industry and the quality of the source text. While this significantly reduces human intervention, our experience shows that human review is still necessary 100% of the time to ensure accuracy and readability. Errors in the source text can lead to translation mistakes, and some machine-generated translations may not be as fluid as those produced entirely by human translators.

The MTPE Process

For our regular or contract clients, we first create a translation memory and a glossary and train the machine translation engine when possible. A human translator with expertise in the relevant industry then reviews, corrects, and approves the translation before finalizing it.

For one-time or walk-in clients who choose MTPE services, we select a machine translation engine profile that best matches the industry of their content and follow the standard MTPE process to ensure the best possible outcome.

What Kind of Content is MTPE Suitable For?

MTPE is ideal for content that is low-risk or low-value but involves a high volume of words, such as:

• Blogs used for search engine optimization (SEO)
• Product descriptions where multiple products have similar descriptions
• Questionnaires with straightforward questions and answers
• Large-scale projects with tight deadlines

As machine translation technology continues to improve, and as clients become more aware of the benefits of MTPE, more businesses are considering this process even for higher-exposure content.

The Benefits of MTPE

• Saves time and reduces costs compared to traditional human translation
• Increases efficiency for high-volume projects with short deadlines
• Provides a scalable solution for companies with ongoing translation needs

Disadvantages of MTPE

• The quality and style of writing may not match that of a human translator
• Machine translation engines struggle with creative or marketing content that requires transcreation

In the translation industry, we use the term ‘Transcreation’ to describe a process where creative or marketing text is not directly translated but rather rewritten to convey the same meaning and emotion in the target language. This approach considers cultural preferences and local market factors, something machine translation engines still struggle with.

If you would like to learn more about MTPE and whether it is suitable for your project, feel free to contact us. We use industry-leading translation technologies, employ trained linguists, and are ISO-certified to provide high-quality MTPE services.