NWS Translates Life-Saving Alerts with Lilt >>
Read More
Accelerate translation with the Lilt Engine
We train our NMT engine on historical translation data and then refine it over time. With every word that our translators type, the Lilt Translation Engine learns and gets smarter - enabling higher accuracy, greater consistency, and increased speed.
The Lilt Translation Engine uses AI and machine learning to analyze existing translation data and make predictive translation suggestions. These suggestions dramatically increase translator accuracy and efficiency by enabling translators to spend their time on new or particularly challenging translation work while the system handles predictable and repetitive parts automatically.
As translators interact with the translation suggestions, the Lilt Translation Engine learns and updates future suggestions based on how they reacted to prior ones. By adapting in real-time, translators see substantial efficiency gains from one sentence to the next. Best of all, the system updates those improved suggestions across the rest of the document as well as all future projects for your organization. That means that each translator is working not just to translate content, but also to train our systems to be more accurate for your organization.
We train the Lilt Translation Engine on as much data as we can find, and we regularly retrain it with new data, languages, and improvements to the predictive models behind the engine. The best part is that each word translated on the Lilt Platform provides valuable feedback that is automatically incorporated to train our systems on your organization's content. But don't worry - your data is completely secure and only used to train your organization's engine. Data is never shared from one organization to the next.
Lilt sets a new standard for AI-powered translation
Your business's content and brand are too valuable to entrust solely to an algorithm with no human supervision. That's why the Lilt Translation Engine, Lilt's adaptive neural machine translation engine, augments rather than replaces human translators.
How does it work? First, we never use post-editing, which is the term for pre-translating an entire document and asking translators to go through and fix mistakes. Instead, our engine makes suggestions to translators as they do their work, helping them to translate faster, more consistently, and more accurately with no loss in quality. If a suggestion isn't accurate, the translator ignores it and continues working on that segment. If a suggestion is accurate, the translator accepts it and moves on. Whichever decision the translator makes, the system has an additional data point on the accuracy of its suggestion, and learns in real time for the future. If similar segments appear elsewhere across that same document or any other existing projects or documents, the system updates those translation suggestions as well.
The adaptive Lilt Translation Engine is the future of language translation, unleashing new levels of human productivity that make it faster, simpler, and more affordable for every organization around the world to translate all of their content in all of the languages spoken by all of their customers, constituents, and stakeholders.