SlatorCon Zurich is now SOLD OUT — See you on October 4th!
News and analysis of the latest developments in machine translation, computer-aided-translation, natural language processing, and other language-related areas in artificial intelligence.
Researchers from NVIDIA, Factored.ai, Talon Voice, and others open-source a properly licensed dataset of 1,780 hours of speech in 77 different languages, plus transcriptions.
Meta introduced SeamlessM4T after months of research and numerous expansive multilingual models. How does the “next big thing” fit into future projects, and the language industry?
Google created a new dataset for machine translation and multilingual NLP tasks across 400 languages and released a high-performing multilingual MT model trained on this data.
G/O Media fires Gizmodo’s entire staff of Spanish writers and begins publishing raw machine-translated versions of English articles that are now being indexed by Google Search.
DeepMind introduces a new method to improve the quality of large language models. Researchers choose machine translation as a use case to show how well the new approach works.
Language AI researchers show that fine-tuning large language models with fine-grained human judgment data boosts machine translation evaluation.
Meta continues to pour resources into multimodal translation capabilities as it launches SeamlessM4T, capable of translating speech and text and trained on hundreds of thousands of hours of data.
Logrus Global and the University of Manchester showcase that fine-tuning LLMs on historical post-editing data can pinpoint segments that require editing and predict translation quality.
ADAPT researchers introduce adaptNMT, an innovative open-source application designed to simplify the development and deployment of machine translation models.
MATEO aims to open up machine translation evaluation, making it accessible to more stakeholders and facilitating research, education, and critical evaluation of machine translation.
Boğaziçi University researchers reveal the potential of customizing MT systems to replicate a translator's style, leading to literary translations that mirror the translator's unique style.
The Post-Edit Me! project supports educators in developing innovative post-editing training practices, fostering enhanced post-editing skills for the future of the industry.
Chinese NLP researchers open source BigTranslate, a large language model based on Meta's LLaMA that its creators say works well for machine translation in low-resource languages.
Google updates privacy policy. Spokesperson clarifies that the latest version not only covers Google Translate and other existing services now but also new AI co-pilots like Bard.
Center for Democracy & Technology examines the outsize impact of English — and machine translation — on multilingual models, and advises action for public and private players.
Researchers at the University of Groningen found that using training data from multiple source languages improved the accuracy of both monolingual and multilingual classifiers.
Google's AudioPaLM fuses text-based PaLM-2 and speech-based AudioLM into a large language model, elevating speech recognition and translation while preserving authentic voice quality.
IndicTrans2 addresses the challenge of machine translation for Indian languages with the researchers providing accessible models, comprehensive training datasets, and diverse benchmarks.
PolyVoice's double language models and smart segmentation along with a decoder-only approach innovates speech-to-speech translation, benefiting unwritten languages.
Microsoft and the Chinese University of Hong Kong study shows enhancing large language models with multilingual dictionaries improves machine translation.
Slator Weekly: Join over 15,600 subscribers and get the latest language industry intelligence every Friday
Tool Box: Join 10,000 subscribers for your monthly linguist technology update.
Your information will not be shared with third parties. No Spam.
This will close in 0 seconds