TAUS

🌟 What an incredible week of connection, collaboration, and creativity! 🌟 The whole TAUS team came together in Amsterdam this week for an intense program of brainstorming, planning and team building. There's something truly special about coming together in person to share ideas, chart our course, and strengthen our bonds as a team. From intense brainstorming sessions sparking innovation and new ideas, to strategic planning to align our vision, and survival games that tested our teamwork, every moment was an opportunity to grow stronger as a unit. The highlight? Without a doubt, it was the chance to spend time together, share laughs and good conversations, and forge deeper connections beyond our day-to-day work. The activities planned throughout the week definitely revealed some hidden talents and common interests we never knew existed among us! As we reflect back on the past few days, we carry with us not just new ideas and strategies, but a renewed sense of camaraderie and shared purpose. The blend of different nationalities, cultures, and viewpoints in our team fuels our creativity and problem-solving, and enriches our company culture. This week has energized us all, reinforcing our commitment to the bright future for TAUS we're building together. With Jaap Van Der Meer, Dace Dzeguze, Anne-Maj van der Meer, Amir Kamran, Simona Beccaletto, Nikos Argyropoulos, Zachary Haynes, Agustin Rossi, Lahorka Nikolovski, Lisa Vasileva, Shikha Sharma, David Koot, Martina Dankovic, Aristotelis Gorgias, Symeon Spyrou, Sofie Kintcheva and George Evgeneiadis #TeamBuilding #Innovation #WorkCulture

From using QE to assess PE efforts and win more business in the pre-sales phase, to using LLMs for multilingual content generation, to ensuring quality and saving costs in multilingual customer support. AI and LLMs are inspiring a lot of new processes and features. Read more about it in our QE buyers Considerations Report where we shared benefits - and showstoppers of buying the Estimation API. If you want to learn more about quality estimation, don’t hesitate to reach out to us directly. https://hubs.ly/Q02Frf_D0

How is Quality Estimation different from Quality Evaluation? 🔍 Quality Evaluation happens after translation, comparing MT output with human translations to assess performance over time. Quality Estimation (QE) occurs in real-time, predicting MT output quality without human references. QE scores measure semantic similarity between source and target segments using advanced language models. Unlike TM matches, which show textual similarity, QE scores focus on semantic similarity without needing references. QE scores are estimates and can vary by context, which is why custom models tailored to specific domains and language pairs provide more accurate results. 📈 Read more in the "Understanding QE Metrics" section by clicking the link below: https://hubs.ly/Q02Frr5F0

On 20 June at the TAUS Massively Multilingual Conference in Rome Sébastien Bratières, Director of AI at Translated, presented an update on the EU-funded project Meetween that brings video conferencing to the next level. Featuring automated meeting secretaries and a virtual agent to represent absentees. This work is done by developing foundation models for speech, which are large neural networks inspired by LLMs. Read more about this 4-year project on the website: https://www.meetween.eu/

Our first #TAUSABQ2024 newsletter is out! Read more about the newest updates on the conference. ⬇

🇺🇸🎆 Happy Fourth of July! 🎇🇺🇸 Nothing planned for the long weekend? Did you know you can try our Quality Estimation API for free? Unlock the benefits of MTQE and improve your translation quality while reducing costs. Get 500,000 characters free of charge and see how our API can streamline your translation workflow. 🚀 https://hubs.ly/Q02FmsxB0 Don't miss out on this opportunity to elevate your translations with TAUS and let us do the work for you. Start your free trial today!

The goal of MTQE? Streamlining your content workflow, enhancing quality, and ultimately saving you time and money. 🚀 But how exactly does it help you? 1️⃣ Applicability: Unlike traditional metrics, MTQE is a standalone powerhouse, making it more scalable and adaptable across diverse language pairs and scenarios. 2️⃣ Data-Driven Automation: MTQE is a fully automated, data-driven process that evolves through iterative training, constantly improving to provide more reliable quality scores. 3️⃣ Beyond Cost Reduction: While MTQE is a perfect solution for reducing post-editing costs, there are other use cases as well. Risk management in high-volume or rapid-delivery scenarios, finding the right MT engine for your language and domain, and enhancing your LQA process with smart sampling. MTQE is a game-changer for efficiency, risk mitigation, and strategic quality analysis in translation workflows. ⚙️📊 Read more about V2 of the TAUS Estimate API, showing increased correlation with human evaluations and stay tuned for further updates on TAUS Quality Estimation and its capabilities. 📈 https://lnkd.in/dQ7GaRCJ

Looking back at #TAUSRome2024, one thing is clear: Quality is still one of the main topics to discuss. One way to get on top of your quality control is through quality estimation. Combining Machine Translation (MT) with Quality Estimation (QE) is the key to achieving faster, more efficient, and cost-effective workflows. 📈 Not sure how it can benefit you? Try out the TAUS Estimate API with our free trial. 🚀 Register now and gain access to our Sandbox mode with a bundle of 500,000 characters to process in 5 languages: English, German, Spanish, French, and Italian. Sign up now for your free trial and discover how to enhance translation efficiency while maintaining quality standards and keeping up with the demand! https://hubs.ly/Q02DyKxf0

Recently we conducted a survey around the buyers' considerations for Quality Estimation which resulted in a comprehensive view of the industry's mixed reactions towards automated quality solutions. 📊 While the benefits of QE—such as increased capacity, faster service, and informed decision-making resulting in cost savings—are acknowledged, significant concerns and uncertainties about Large Language Models persist. Clearer communication about QE’s capabilities, robust use cases, and simplified scoring metrics are crucial. Check out this latest TAUS report, consolidating the results from the survey, as well as feedback from the QE Workshop in Rome last week, providing a clear overview of the current role of quality estimation in our industry. 📚 https://hubs.ly/Q02DymWt0

The dust has settled and we are sharing the last newsletter on our #TAUSRome2024 Conference. 🇮🇹 Thanks to everyone who was a part of this and we will see you in October at #TAUSABQ2024! 🚀