Issue Focus

NCSoft has launched an advanced AI model called 'Zero shot Multiverse TTS,' designed to revolutionize game voice production. The Text-to-Speech (TTS) system can generate character voice content directly from written input. It features abilities like diverse speech style conversion, exceptional speaker timbre matching for cross-language generation, and the capability to produce multilingual voices using just a three-second audio sample.
The company intends to implement 'Multiverse TTS' widely across its game development processes, aiming to significantly cut down on the time and expenses traditionally needed for voice work. This technology allows the creation of high-quality AI character voices using minimal voice data, thereby boosting production efficiency.
A standout characteristic of 'Multiverse TTS' is its ability to operate as a single model supporting multiple languages and functionalities, offering a cost-efficient solution for quality voice generation. To underline their technological prowess, NCSoft has presented a paper on this model at the esteemed AI technology conference, EMNLP.