If you’ve read any articles I’ve penned, it’s clear that I’m a bona fide, linguistics-for-days language and storytelling nerd. Just always been this way, it would seem.
That said, when I see the word “lilt,” for example, my brain goes to the diverse cadences of folks’ speech patterns, what makes how they intonate unique, the visuals they create when telling stories, how potential story excitement affects their voice, all that jazz. To many – myself obviously included – the differences and how they’re generated are fascinating.
In similar words, what’s described here is essentially the lilts with which people speak; the almost-musical emphasis on vocality and what’s being effectively conveyed (and even how that impacts how those same people write and text). It’s a very human thing; not to sound too E.T. here, but that’s still noteworthy because it isn’t terribly often that inventions like AI are capable of replicating it. (Or, if they indeed are capable, it’s usually neither as passable nor as satisfyingly convincing as how we humans naturally do it, ourselves.)
However, not everything language-wise must be completed manually. Our technology can help.
Enter Lilt. Trusted by the likes of Intel, Canva and Juniper Networks (to name a few), Lilt was founded by researchers from Stanford and Berkeley, where each had worked on language technologies and sought to make information more accessible to everyone. Once these two researchers met while working on Google Translate at Google, they got together and decided to build a company of their own. (This was also in 2015, i.e. before machine-learning (ML) capabilities and overall machine translation didn’t quite meet the higher-quality standards they do now; especially before full enterprise translation services came to the fore.)
So, Lilt has prioritized its investments in large language models (LLMs) to best create a way that everyone can access available information in their preferred language. That’s also key; not only the speech-related nuts and bolts, but the languages spoken. This is another core element to what should be a modern-day given, in terms of accessibility.
Thus, in addition to Lilt’s existing Contextual AI Engine and Platform – which learns from real-time feedback from linguists, enabling high-quality on-brand, multilingual translation – Lilt also recently announced the launch of a new tool:
Lilt Create.
With Lilt Create, enterprises can generate instant, on-brand content – in any language. Having been trained thoroughly on linguistic-heavy datasets (and able to be trained on data from companies’ databases and brands’ campaign contexts), With Lilt Create, its generative AI is capable of developing locally relevant, enterprise-grade multilingual content that is worded well (as most humans do) and is delivered much more quickly, reducing employee downtime. Via self-learning models, as well, Lilt Create rapidly understands and identifies users’ prompts, produces results in the language(s) desired, and can even help brands render contextually relevant style guides, based on brand conversations and other iterations of content.
“Historically,” said Lilt CEO Spence Green, “enterprise organizations that operate in global markets have been forced to choose between creating source content all in one hub – and then translating it to other languages, typically manually – or instead, decentralizing content creation.”
“But now,” Green continued, “those organizations need not worry about centralization versus decentralization when it comes to global content creation strategies. Lilt Create provides the answer to the speed-versus-local-relevance trade-off. It offers customers a powerful tool to empower their teams and develop optimized, fit-to-purpose strategies that are more accessible, more appropriate for their needs, and more befitting of the tones and styles that customers engage best with.”
Lilt Create is available in public preview for enterprise users through July 2023. After that, we’ll have to see where this advanced AI solution for translation and localization goes next, in regard to solving businesses’ language and content-related challenges.
Edited by
Greg Tavarez
