À̹ÌÁö È®´ë SK C&C under South Korean conglomerate SK Group whose flagship is top wireless carrier SK Telecom claimed it has commercialized artificial intelligence technology to recreate human speech based on taped voice clips.
A 20-minute high-quality recording of a target voice is sufficient to analyze and recreate the characteristics of the speaker`s voice, tone, intonation, pronunciation, and speed, SK C&C said.
Unlike the existing text-to-speech technology, which divides the voice into consonants and vowels and then recombines them to read sentences, the new AI technology analyzes and learns the speaker¡¯s speech corpus, which is a database of speech audio files and text transcriptions.
Previously, it took more than a month to produce an audiobook with a human narrator, but the company said that this technology enables to create a voice of celebs or voice contributors using their pre-recorded voice clips, contributing significantly to reducing the entire production period.
SK has worked with voice-cloning startup LionRocket as well as the National Research Institute of Cultural Heritage to support the production of narrative descriptions in a braille tactile book for the visually impaired to introduce natural monuments. The voices come from Korean actors Lee Byung-hun and Han Ji-min who expressed their intention to donate voices for the visually impaired.
The two who signed a business agreement plan to explore cultural and educational content markets initially with for potential applications of this technology.
Other projects being planned include global content business using Hallyu, or K-wave, stars and characters, and businesses such as reading books for adult consumers and children.
LionRocket will also explore the global market by developing a function that automatically generates foreign languages such as Chinese and English based on only Korean voice data, said the company¡¯s CEO Jung Seung-hwan.
By Pulse
[¨Ï Pulse by Maeil Business Newspaper & mk.co.kr, All rights reserved]