Hancom WITH’s ‘SPEEKEY’ Proves Its Voice Authentication and Deepfake Detection Capabilities Globally

[Financial News] The underlying technology of Hancom AISRC SPEEKEY, Hancom WITH’s deepfake-detection speaker-authentication solution, has received recognition at the world’s most prestigious speech and signal-processing conferences. Hancom WITH is the de facto holding company of Hancom Group.
According to Hancom Group on the 11th, the Soongsil University AI Security Research Center, which conducts research and development on SPEEKEY’s core technologies, achieved notable research results this year at Interspeech and ICASSP, two of the world’s leading conferences in speech and signal processing. The validation results are regarded as a reliable measure of the commercial model’s credibility, going beyond the establishment of theoretical foundations.
Seven papers were accepted at Interspeech this year. This is an achievement that is difficult for a single research institution to accomplish in one year, demonstrating globally competitive capabilities across various core technology areas, including speech recognition, speaker verification, and synthetic-speech detection.
At the ICASSP WildSpoof Challenge, the center ranked second globally. WildSpoof Challenge is an international competition in which technologies for generating synthetic voice deepfakes and detecting them are tested using real-world, or “in-the-wild,” data. It aims to address real-life speech conditions, including noise and outdoor environments, overcoming the limitations of previous evaluations that relied on studio recordings without background noise.
In particular, in the SpoofCeleb evaluation, which reflects actual everyday noise and diverse recording environments, the center recorded the highest performance, surpassing the overall top-ranked team. The result demonstrated its ability to accurately distinguish fake voices in real-world environments containing call noise and everyday background sounds.
Based on these research achievements, Hancom WITH plans to improve SPEEKEY’s accuracy in detecting AI-generated synthetic voices and expand its business into industries requiring high levels of security, including voice authentication in the financial sector and voice-phishing detection at call centers.
Sangyeop Song, CEO of Hancom WITH, said, "These achievements demonstrate that SPEEKEY’s technical reliability and differentiation are at a global level." He added, "Based on its proven technological capabilities, we will continue to strengthen SPEEKEY’s competitiveness and establish ourselves as a leading company in the next-generation security market, protecting customers from cybercrime that exploits generative AI."

[email protected] Yeon Ji-an Reporter