K-Content News

Expressing Emotions by Voice, AI Voice Actor “Typecast” Kim Tae-soo, CEO of Neosapience
  • August 12, 2020

Expressing Emotions by Voice
AI Voice Actor “Typecast”

Kim Tae-soo, CEO of Neosapience

This is certainly the age of artificial intelligence (AI). Mixing AI with diverse cultures is emerging as a new icon. Can AI replace voices? Neosapience, a speech synthesis technology startup, received attention by unveiling its AI voice actor service “Typecast.” Voices filled with souls and emotions are applied in various fields to show abundancy in contents. We met CEO Kim Tae Soo who is making remarkable moves with his incomparable technical prowess.
(위드코카2 사진2)
Creating voices that comfort the listeners
“I Met You,” a documentary film screened on MBC in February, stole the limelight by making what’s impossible a reality through the VR content.It gained recognition by reenacting the appearance of a daughter who died of incurable disease and showing scenes of reunion with her mother. The child’s appearance was restored by analyzing her facial expressions, gestures and voice based on her photography and video data. The child made viewers move to tears during the scenes when she looked for her mother and when she asked her mother if she had been thinking of her. It is Neosapience that restored the voice of the deceased child. CEO Kim Tae Soo said, “The production staff from MBC visited me last fall, asking for voice restoration. They showed me a video of the child’s lifetime appearance. But I was worried at first. It was a one-minute video but the child didn’t have a good command of sentences. Apart from the difficulty in voice restoration, I was worried that there might be a possibility of ethical controversy.”

Kim took part in the production with a bit of hope that the program’s good intention will be relived. Lack of data was completed after going through deep-learning processes following the dubbing of more than 800 sentences each with the voices of her peers. It was the result of Neosapience’s unique technical prowess plus the enthusiasm of professionals.

(위드코카3 사진1)
From possibility to reality; from expectation to opportunity

It was in 2017 that Kim founded Neosapience. Having earned a doctor’s degree at the Korea Advanced Institute of Science and Technology (KAIST), he made a fresh move with his co-workers and juniors in KAIST laboratories while working on audio machine learning at LG Electronics and Qualcomm. They began to study the innovative speech synthesis technology with the aim of changing the world through their peculiar AI technology. Of course, things were not easy from the beginning. Human voices show different meanings according to nuances and subtle changes in speech. Kim devoted himself to the research, questioning himself whether it’s possible to reenact the characteristics of each voice with AI technology.

He buried himself in the research to make the small chance into a reality. Consequently, he succeeded in realizing voices expressing human feelings beyond merely delivering voices clearly and mechanically. As its first experiment, Neosapience created a video synthesizing the voice of President Trump in Korean language ahead of the inter-Korean summit in 2018. The video went viral at famous overseas sites and gained global recognition. The company drew more attention after unveiling the video of North Korean leader Kim Jong-un speaking in English.

“I believe we were able to win far more attention because our research was the world’s first ever. It was very meaningful that no one dared to attempt this kind of project. This idea inspired us to work more and develop more advanced technology,” he said.

(위드코카2 사진3)
Emotion-filled voices captivate the hearts of users

Neosapience unveiled its premier AI voice actor service Typecast last year. Typecast is a service that transforms texts into audio contents based on AI speech generating technology, using the voices of professional voice actors. If the user selects a voice actor according to gender, age and content atmosphere after entering the sentences, he or she can download the audio files as if recorded by a professional.

“Neosapience is the first company that developed the speech synthesis source technology expressing emotions and characteristics. In fact, Text-to-Speech (TTS) service was first introduced 30 years ago. The subway announcement and robot voices in the movie Star Wars are some of the examples of TTS. Apple’s Siri and smart speakers belong to the conventional technology. Yet, I think that the sophisticated speech synthesis technology that can express delicate feelings and adding cadence by using the voices of professional voice actors is our unique competitive edge.”

Typecast services are used broadly in diverse fields such as media and entertainment based on its unique technical expertise. They charmed the industry as well as users for documentary films, broadcasts, video dubbing, news production, audiobooks and educational contents.The members have exceeded 30,000 only seven months after its launch.

He said, “I feel rewarding when I see users creating satisfactory contents with Typecast. Our services are now a tool to create various contents. I feel a sense of satisfaction after discovering that our services were of help to higher-quality contents.”

Evolution of technology continues

Neosapience is widening its spectrum by making another challenge. It has been taking part in an R&D project of the Korea Creative Content Agency (KOCCA) since last year. The company joined the project supervised by Soo-Young Lee, director of KAIST AI Center, who is counted among the country’s famed AI experts, and dedicated himself to the technological development of emotional audiobooks. He noted, “Audiobooks or ‘hearing books’ are becoming increasingly popular worldwide. Especially in abroad, demand for audiobooks is high. We developed a technology that reads books with emotion-filled voices rather than conventional machine sound. We already signed a partnership contract with Daekyo last year and completed two audiobooks. Such earlier-than-expected results are encouraging. We’ll have to exert more effort to complete the rest of our research by the end of the year.”

Kim says there is a long way to go. Given that improvements and complementary matters are being discovered, depending on users with various needs, he knows that technological development is always ongoing. Aside from raising the technological quality, he expressed his will to listen carefully to the smallest suggestions from users and create more effective technology by accommodating diverse needs. He continued, “I am often reminded of our goal when the company was founded. I wanted to do things that could give a positive impact on people around the world. As to technology or services that are nestled deeply in our lives, we frequently say, ‘That’s too complicated. You can simply use this service.’ I have a dream and objective of hearing in the distant future that ‘That’s a complicated way of creating content. Make it simpler by using Neosapience’s services.’”

Neosapience contributes to raising the degree of completion of contents with unprecedented new technology. One cannot but wonder about their next move to witness how they will amaze the world once again.