NC Voice Synthesis Technology Presented at International Conference Paper

Boasting Perfect Sound Quality Hard to Distinguish from Real Voice

[Asia Economy Reporter Jin-gyu Lee] NCSoft (hereinafter NC) announced on the 23rd that its voice synthesis technology 'VocGAN' will be officially presented at 'Interspeech 2020' in October.

Interspeech is known as the world's most prestigious international conference in the field of speech recognition and synthesis.

VocGAN is a neural vocoder technology independently developed by the Speech Synthesis Team of the 'Speech AI Lab' under NC's Artificial Intelligence (AI) Center. It features perfect sound quality that makes it difficult to distinguish between recorded and real voices, fast speed, and lightweight design.

NC is applying voice synthesis technology in various fields. The narration of the tutorial video 'Guide' for the Frontier World of 'Blade & Soul' was entirely produced using synthesized voices. This eliminated the process of voice actors learning lines, recording, and reviewing, thereby shortening production time.

Voice synthesis technology is also applied in the game development process. It is used in optimization testing stages such as reviewing the content and volume of character dialogues during development. NC also operated 'NC AI Books,' which introduces a book using AI synthesized voices through the company's internal broadcasting. An NC official stated, "We have automated the entire process from voice data accumulation, deep learning model training, synthesized voice generation testing, to final service on the cloud."