AI Solutions

AI or artificial intelligence

Machines in our lives

Computer assisted translation (CAT) tools have been available since the late ’90s to aid in the translation process by storing previous translations in a memory. Machine translation appeared in the field later, and now there are also large AI-based language models, or neural machine translation tools, that can make sense of what is being translated. The result, however, will not be of the highest quality without a human touch. Even so, these tools already do quite a good job of understanding the text.

CAT tools (such as MemoQ) find an exact or close match from the translation memory, but a machine translation tool that has been trained to think based on previous translations (which, incidentally, is a very expensive process) can also provide logical translations for non-exact matches. The quality of translations into more common languages is better than translations into Estonian. However, translation memories expand every year and the final result continues to improve over time.

So what is the point of a translation agency if computers are already so smart?

We continue to translate in the traditional way and use technological tools to support us in our work. You can’t rely entirely on machines: it is the human hand and eye that humanise translation and make it logical. Localisation plays a huge part in this.

AI services at Luisa Translation Agency

Machine translation and post-editing

Machine translation and AI-based language model technologies use sophisticated algorithms and neural networks to translate texts quickly and in large volumes. Machine translation is a great tool for obtaining preliminary translations in certain languages and on certain subjects. Translators and editors then review the result to make sure the final text is accurate and sounds natural.

We offer (in addition to standard translations, of course) both machine translations for a quick overview of high-volume texts and machine translations with post-editing.

When post-editing, the translator checks the text against the original. The result is equivalent to a standard translation.

Note: Although machine translation tools are expensive, machine translation is still cheaper than standard translation.

ChatGPT

ChatGPT is a well-known tool that helps clients compile and translate their texts. We can also help you improve these texts through editing and localisation. We are getting more and more requests for adapting and/or translating texts created by AI.

Why is a machine translation from Luisa better than one from Google Translate, for example?

Google Translate and similar machine translation tools (such as DeepL) have come a long way in recent years, and everyone has likely used Google Translate at some point in their lives. Most of the world’s translators and translation agencies use machine translation tools as supporting software to some extent. In addition to our own machine translation tool (see below), we use paid licences for other translation engines, which guarantees better results than the free versions that are available online. What is most important, however, is that your texts and data remain confidential and do not leak onto the internet.

In cooperation with the University of Tartu, Luisa has developed its own machine translation tool based on the millions of pages of translations we have produced over the past 30 years in our main language combinations. This intelligent neural machine is trained to understand the source text and provide meaningful translations based on our translated and edited texts.

It should also be noted that all these technical solutions are a huge investment, but we make them with the aim of achieving high-quality texts and fast results in cooperation with humans (translator + editor). We provide translation services in over 200 language combinations. In almost 80% of them it is impossible to receive a high-quality translation with these tools alone.

• A fast and high-quality solution that is suitable for many fields
• A human translator makes the final text accurate and sound natural.

Speech recognition and transcription

AI-based speech recognition tools are already rather good at converting audio files into text, which can be used to document a wide range of verbally presented material. Again, a translator and editor help add the final touches.

Multimedia localisation

AI technology is already capable of assisting in the localisation of multimedia content, such as videos and audio files. This includes automatic subtitling and translation and audio and video synchronisation by target language. The final human touches ensure high-quality results.

We have 30,000 corporate and government clients and over 5,000 private clients