MERaLiON: Multimodal Empathetic Reasoning and Learning in One Network
Human-Centric AI for Southeast Asian Speech and Beyond
Multimodal Empathetic Reasoning and Learning in One Network (MERaLiON), developed by A*STAR Institute for Infocomm Research, is an advanced large language model that captures how Southeast Asians truly communicate. It goes beyond words to include tone, speaker identity, and cultural nuances such as code-switching and local expressions.
Unlike conventional systems that rely on converting speech to text, MERaLiON processes speech directly, preserving vital non-verbal cues. This leads to more natural, accurate, and humanlike AI interactions, with applications spanning speech translation, customer service analytics, and voice-based AI wellness tools.
As part of Singapore’s $70M National Multimodal Large Language Model Programme, MERaLiON is open source and industry-backed by ST Engineering, Grab, Singtel, and Microsoft, driving real-world adoption across the region.
Deepfaic: Safeguarding Online Trust and Safety
Trustworthy AI for the Age of Deepfakes
Deepfaic, a spin-off from A*STAR Institute for Infocomm Research, developed an AI platform for detecting deepfakes across video, audio, image and text inputs. Trained on more than seven million examples, Deepfaic integrates into enterprise workflows for real time verification. It supports secure video interviews and online Know Your Customer checks, protects organisations from impersonation and fraud, and helps publishers validate content authenticity.
Backed by the Centre for Advanced Technologies for Online Safety, Deepfaic is bringing trusted AI tools to enterprises and institutions, safeguarding brands and public discourse in an increasingly AI driven world.
