NC AI announced on the 17th that it will unveil its monster sound generation and transformation artificial intelligence (AI) technology to a global audience at INTERSPEECH 2025, the world’s largest conference on speech and language technology.
INTERSPEECH, organized by the International Speech Communication Association (ISCA), is the world’s largest conference in the field of speech and language technology. Every year, speech researchers and industry professionals from around the globe gather to share the latest research findings and innovative technologies.
The 26th edition of the conference, held from August 17 to 21 in Rotterdam, the Netherlands, is themed “Fair and Inclusive Speech Science and Technology.”
At this year’s conference, NC AI will present two papers: one detailing the architecture and training methods of its high-quality timbre transformation model specialized for monster sounds, and another describing the implementation of this technology as a web-based real-time transformation system.
On site, visitors can participate in an interactive demo where their speech or uploaded sounds are instantly transformed into the cries or roars of specific monsters. An online demo page will also be available, allowing those who cannot attend in person to experience this advanced technology.
NC AI described this technology as a breakthrough that will revolutionize the way monster sounds are produced in large-scale massively multiplayer online role-playing games (MMORPGs). The system analyzes audio at CD quality (44.1kHz), capturing everything from a character’s unique rough breathing to sharp roars, and then overlays only the desired style while preserving the original content of the voice.
With NC AI’s advanced model, sound designers can now expand the broad frequency spectrum of human voices to precisely reproduce the dynamic and complex timbres and textures unique to monsters. This significantly reduces the time and cost previously required for manually creating variations for each monster and situation.
Additionally, the system allows for fine-tuned control over style attributes that reflect character traits such as aggression, intimidation, or playfulness. As a result, even the same monster can automatically generate new sounds according to different combat or emotional states.
The foundation of this technology is a vast repository of high-quality data. The NC AI Audio AI Team, in collaboration with the NCSOFT Sound Center, has meticulously classified and tagged a large-scale game audio database accumulated over many years, further segmenting it by various acoustic characteristics such as timbre, airiness, noise, and mood.
Cho Namhyun, Head of the NC AI Audio AI Team, stated, “As a leading research organization in Korea’s multimodal AI sector, NC AI has completed this monster sound transformation technology by combining a massive game audio dataset, cutting-edge AI modeling, and outstanding sound design expertise. Moving forward, we will continue to leverage AI to turn creators’ imaginations into reality and deliver innovative audio experiences across the entire digital content industry.”
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
