NC, Paper Published at AI's Largest Academic Conference 'ICASSP'

NCSoft announced on the 18th that it has published a total of four papers at 'ICASSP (International Conference on Acoustics, Speech and Signal Processing) 2024,' the largest international conference in the field of artificial intelligence (AI).

ICASSP is an international conference held annually in various countries around the world. This year marks its 49th edition and will be held at COEX in Seoul. The event will host over 4,000 scientists worldwide who will present various multimodal generative AI technologies, including signal processing techniques.

At ‘ICASSP 2024,’ NCSoft published four papers on ▲mitigating visual hallucination phenomena in multimodal language models ▲improving facial recognition accuracy in harsh environments ▲wake word recognition models that cleanly filter user speech ▲virtual voice generation systems matching facial images.

NCSoft stated, "The papers presented this time are meaningful in that they address the potential applications of ‘AI applied technologies’ for practical use, as well as the utilization possibilities of ‘multimodal language model technologies’ that understand and learn various data such as text, images, video, and speech."

Based on these research results, NCSoft plans to develop AI technologies that understand multimodal data across various fields such as text, images, and audio, and apply them to game development.