Historically, translators have always relied on the translation memory – popularly referred to as the heart of the CAT tool – for productivity. When contemplating machine translation (MT) as either an alternative or a complement to the CAT tool, the traditional consensus had been that MT technology was not yet accurate or intuitive enough to be an efficient tool in this time-conscious industry.
This mentality started to shift in 2016 however, as the advent of neural machine translation (NMT) came to the fore and its impressive capabilities began to capture the industry’s attention. As part of our commitment to providing secure, high quality machine translation, the SDL development team had ardently been working towards the creation of a sophisticated neural offering, and towards the end of 2019, they proudly announced that our very own NMT engines were now what powered SDL Machine Translation.
Many Studio users will already be familiar with accessing SDL Machine Translation (SDL’s machine translation functionality) through the Language Cloud interface, but behind the scenes, our development team had been busy transitioning from the previous statistical model to the powerful, home-grown NMT model.
From its introduction in October 2019, we have offered 500,000 free NMT characters per account, per month for our Studio users and the response has been remarkable!
In October alone, there was a 250% growth in users utilizing SDL Machine Translation, and usage further continued to increase by an average of 75% each consecutive month. Studio supports over 40 different machine translation engines and yet within a matter of weeks, our own offering was quickly accelerating through the ranks in popularity.
Our NMT has not only been so well received because it has a free monthly usage, but also because of the high quality results across a wide range of language pairs. To highlight a specific example, last year, our research team announced a breakthrough had been made in the development of our NMT engine, and when the Russian to English machine translation system quality was measured, it outperformed all industry standards. Over 90% of the system’s output was labelled as perfect translation by professional Russian-English translators, setting a new benchmark for Russian to English Machine translation. Click here to read the full article.
Another reason for its soaring popularity could well be that SDL is regarded as a trusted source, providing secure machine translation services. “SDL does not retain your data,” Daniel Brockmann, Director of Product Management at SDL explains, “and this is a stark contrast to some other cloud providers who state in their terms and conditions that they capture any data that comes in for various purposes.”
At the time of writing this blog, I can confirm that SDL Machine Translation has now achieved the accolade of being the most frequently used machine translation engine in Studio, despite there being a plethora of other machine translation engines supported by our software.
“No one else can match this flexibility,” Daniel further remarks when addressing the ability to choose between over 40 different MT providers, including DeepL and Google which have now been relegated to second and third place. “So I would say that SDL Trados Studio can now be seen as the ‘home for neural machine translation.’ - This is an ambition that I have always been passionate about.”
“SDL Machine Translation was built to serve the needs of anyone that wants security, speed, quality and ease of use.” Jane Hendricks, Senior Product Marketing Manager at SDL adds. “The billion characters used through SDL Trados Studio demonstrates that it can meet the needs of translators, alongside enterprise business functions like customer service, business intelligence, eDiscovery and internal communication. SDL Machine Translation does more than deliver word-in/word-out capabilities, it is able to fit into multiple workflows easily and break language barriers wherever they exist.”
To learn more about SDL Machine Translation, and how it can help you process much larger volumes of work whilst maintaining the high quality demanded from your customers, watch Daniel Brockmann and Neil Ferguson’s introduction to NMT webcast.