NC AI announced on the 17th that it will unveil its innovative monster sound generation and transformation technology at the world's largest speech and language conference, INTERSPEECH 2025.
The conference will be held from Aug. 17 to 21 in Rotterdam, Netherlands, and NC AI will present two papers related to a high-quality sound transformation model specifically for monster sounds and a real-time web demo using it. At the experience zone, visitors will also be able to see demonstrations where speaking into a microphone is instantly transformed into the cries or roars of specific monsters.
The core of NC AI technology is to apply the desired style while preserving the content of the original voice. It analyzes voices at CD quality (44.1 kHz) to meticulously reproduce textures such as rough breathing and metallic resonance, reflecting intensity and rhythm in 0.005-second units to create sounds that resemble those produced by actual living beings.
Through this, sound designers can automatically generate a variety of monster sounds that were previously crafted individually. It also makes it easy to implement various sound variations that reflect character traits such as aggression and intimidation, greatly enhancing production efficiency and creativity.
The foundation of the technology is a large-scale game audio databases built in collaboration with NCSoft sound center. Utilizing specialized sound transformation tools, extreme and diverse non-human tones have also been learned through virtual data augmentation techniques. As a result, the NC AI model surpassed the latest sound transformation models, proving excellence in all evaluation criteria, including sound quality, naturalness, and content preservation.
This technology will also be applied as the engine for NC AI's generative SFX production tool "Sound Palette," and it will be used not only in games but also in the production of various digital content such as films, advertisements, XR, and metaverse. Through this achievement, NC AI plans to strengthen its status as a representative multimodal AI corporations in Korea and accelerate its entry into the global audio AI market.
Cho Nam-hyun, Head of Team at NC AI, noted, "We completed the monster sound transformation technology by combining vast data, advanced modeling, and expertise in sound design."