With the situation unfolding in Europe, we want to lead off this month by highlighting our latest initiative to assist Ukrainian refugees throughout the EU who are in urgent need of translation services. If you require any document translation, including scans and photos (or know someone who might), we’ll grant you free access to our Translation Portal with Character Recognition. Please, register at https://inten.to and write us as firstname.lastname@example.org with UKRAINE in the subject line 🇺🇦
As usual, we also have a few updates reflecting on February to help you grow your business throughout the next month. March will be a time to focus on expanding and improving your language outreach with new language pairs and key updates to data cleaning and metric options in the MT Studio. Make sure to finish Q1 strong with the tips below. Enjoy!
🦾 Three new MT Studio features to streamline your workflow
🗣 More language options added to the Intento platform
🚀 ‘The Elevation of the Localization Manager’ at SlatorCon
🏆 Data-driven MT strategies for greater business value
🤖 The latest GPT-3-like model promises greater accessibility to safe AI
📚 Great reads from throughout the community you won’t want to miss
Enhanced data cleaning, more metric options, and efficient evaluations in the MT Studio
The Intento MT Studio has been improved with a new data cleaning tool simplifying the most technically difficult and time-consuming aspects of data prep to an all-inclusive, easy-to-use tool. At the same time, COMET-QE has been introduced as a new metric working with or without a human reference. All MT Studio features have been integrated into a more efficient workflow — bringing the entire process into a single interface.
Read the full overview here.
More language options added to the Intento platform
These additions fit with our prediction that 2022 would bring an increase of niche, low-resource languages. This is great news, as more languages mean more direct translations between language pairs — not just through English.
Here’s what’s new:
- Upper Sorbian, Germany (hsb)
- Chechen (ce)
- Central Kurdish (ckb)
- Machame (jmc)
- Uigur (ug)
Check out the full list of available languages here.
‘The Elevation of the Localization Manager’ at SlatorCon March 2022
In the e-whirlwind of the last couple of years, we’ve seen global content volumes skyrocket, global teams scattered to the far corners of the earth, and CEOs using terms like “massively multilingual” and “AI-driven global content strategy”. In the eye of the storm sits the loc manager, more strategic and visible than ever, and in a position of elevated importance to the global enterprise. In this panel, we’ll bring together loc managers from various industries to discuss:
- How their role is evolving
- What they are doing to stay ahead of the curve
- What they see on the horizon for the Localization industry
Data-driven MT strategies for greater business value
Mass digitization across the landscape has changed the way most businesses approach MT, with new stock models, progressive innovations in speech technologies, and more customization options available within Enterprise. Our latest article reflects on these changes, drawing insights from localization experts at Adobe, Nimdzi, and E2F to find out the different roles data plays when developing your MT strategy.
Read the full article here.
The latest GPT-3-like model promises greater access to safe AI
EleutherAI has recently released the largest publicly available GPT-3 model (20 billion parameter) GPT-NeoX-20B is already available for download and will likely prove to be very helpful for research purposes. The dramatic increase in size will hopefully encourage accessibility towards broader use-cases and will aid the research towards the safe use of AI systems. Try it out for a look into the future of content generation.
Looking for further reading?
Here are some more updates from the AI/MT community:
- Teaching AI to translate 100s of spoken and written languages in realtime
- German interpreter breaks into tears while interpreting for Zelensky
- A dedicated corpus and evaluation framework for text anonymization
- Machine dubbing startup raises 20M USD
- Red teaming language models with language models
- Inside the dying art of subtitling
- Localization companies nix carbon in the wake of climate crisis