Speech Translation Systems Fall Short on Prosody, Apple Researchers Find
Apple researchers find that speech translation systems struggle with prosody, affecting accuracy, and recommend improvements...
Maria Stasimioti
Array
(
[post_type] => post
[posts_per_page] => 12
[paged] => 18
[post_status] => publish
[author] => 2795
)
Apple and Adobe Researchers Tackle Cultural Nuances in AI Translation
Apple and Adobe researchers introduce a machine translation model that dynamically retrieves culturally relevant terms...
Google Says There’s a Better Way to Create High-Quality Training Data for AI Translation
Google research finds that human-refined AI translations or LLM-enhanced human translations may serve as alternatives...
Researchers Open Source Toolkit to Make AI Translation More ‘Explainable’
Researchers introduce an evaluation toolkit for machine translation, integrating fine-grained error analysis, explainability, and a...
New Real-Time AI Translation Method Draws Inspiration from Simultaneous Interpreters
Research proposes a new method that allows simultaneous machine translation models to adapt their read/write...
Large Language Models No Match for Humans in Literary Translation, Study Finds
Research shows that in the era of large language models, literary translation remains “an exclusive...
AI Translated Benchmarks Can Reliably Assess LLM Performance, Study Finds
The OpenGPT-X Team released machine-translated versions of five benchmarks in 20 European languages for consistent...
Microsoft Re-Interprets Neural Machine Translation Models as Language Models
Microsoft researchers bring traditional neural machine translation models closer to the capabilities of large language...
With MetricX-24 Google Presents Latest Machine Translation Evaluation Metric
Google presents MetricX-24, its latest machine translation evaluation metric, which demonstrates significant performance improvements over...
Google Finds ‘Refusal to Translate’ Most Common Form of LLM Verbosity
Google researchers reveal that verbose large language models, which offer multiple translations or refuse to...
Microsoft and Johns Hopkins Unveil Multilingual AI Translation Model for 50 Languages
Johns Hopkins and Microsoft researchers present a new LLM-based multilingual translation model. The team calls...