[AI Revolution] ③ Adding Emotions to Virtual Human Voices... Neo Sapiens

CEO Kim Taesu of NeoSapiens Founded NeoSapiens
Converts Text into Emotionally Expressive Voice... Realized as Virtual Human
Collaborates with Celebrities like 'AI Juhyunyoung'... Used in Games, Corporations, and Public Offices
"Synergy through Integration with Generative AI like ChatGPT"

Kim Tae-su, CEO of Neosapiens.

There is a YouTube channel called '1 Minute Cooking Ttukttaki Hyung.' An uncle who speaks in the Gyeongsang-do dialect introduces cooking methods while conversing with his elementary school-aged nephew. It has been so popular that it ranked first in domestic YouTube Shorts views for two consecutive years from 2021 to last year. The channel has as many as 2.26 million subscribers. However, few people know that the voices of the two characters appearing on this channel were created not by real people but by artificial intelligence (AI). Even those who were confident they could easily recognize AI voices only realized after subscribing to the channel for quite some time that the voices were virtual, leaving them feeling disillusioned.

Neosapience: Imbuing Human Emotion into Voices

The company that implemented the voices of the main characters on this YouTube channel is Neosapience. Neosapience offers 'Typecast,' a service that converts text into audio using AI voice generation technology. What sets Neosapience apart from other companies is that it does not simply perform mechanical conversion but imbues voices with human emotions. For example, for the text 'Hello,' users can select emotions such as 'happily,' 'whispering,' or 'sarcastically' to obtain the desired voice. Additionally, by utilizing AI virtual actors, it is possible to produce video content as well.

Neo Sapiens collaborated with three AI startups to create 'AI Juhyunyoung'. (Photo by Typecast, YouTube capture)

Last year, Neosapience collaborated with three AI startups to introduce 'AI Juhyunyoung,' which caused a big sensation. Juhyunyoung is a popular MZ generation (Millennial + Gen Z) trending actress known for various segments such as 'Reporter Ju' and 'MZ Office' on Coupang Play's 'SNL Korea.' AI Juhyunyoung is a chatbot that closely resembles the real Juhyunyoung’s appearance and manner of speaking, enabling real-time communication. When corporate or government press releases are input, it reads them out in Juhyunyoung’s unique humorous and lively tone.

As of this month, Typecast has 1.4 million subscribers. Paid services are used in 76 countries. There are about 370 virtual actor characters registered on Typecast. It is highly popular not only among corporations and government agencies but also among those who want to create so-called 'buca' (secondary characters) online to live a second life.

Expectations for Changes Brought by ChatGPT

Avatar Camilla, which Neosapience has been providing on its YouTube channel since the 16th of last month. Camilla is an avatar that integrates Neosapience's virtual human technology with ChatGPT, communicating with users in real time and broadcasting live 24 hours a day. (Photo by Typecast, YouTube capture)

Neosapience is one of the companies most welcoming the changes that the generative AI boom, such as ChatGPT, will bring. This is because the potential uses are limitless when generative AI is combined with Neosapience’s services.

Since the 16th of last month, Neosapience has been offering 'Camilla,' an avatar that responds in real time by linking ChatGPT services to its YouTube channel. Camilla is a kind of virtual YouTuber who plays requested music and conducts 24-hour counseling broadcasts, answering various questions in real time. This is the first service combining ChatGPT with Neosapience’s voice technology. Kim Taesu, CEO of Neosapience, said, "Voice generation technology has made remarkable progress since using deep learning about four years ago," adding, "Generative AI will bring innovation not only in voice and video but also in virtual humans that generate facial expressions and gestures through avatar synthesis technology."

"The Popularization of Virtual Humans Is Not Far Off"

Kim Tae-su, CEO of Neosapiens.

Neosapience was founded by CEO Kim in November 2017. Kim majored in voice and multimedia signal processing using machine learning during his graduate studies at KAIST. He developed products utilizing technologies such as voice recognition, sound quality enhancement, and sound recognition for over ten years at corporate research centers including LG Electronics and Qualcomm.

At the time of the company's founding, there were only two employees, but it has grown into a company with about 60 employees. It completed its Series B investment last year. The total accumulated investment amount is 31.8 billion KRW. Investors include BRV, an Asia growth investment platform of BlueRun Ventures in Silicon Valley, USA, as well as Woori Technology Investment, Company K, and Albatross.

Neosapience plans to actively target overseas markets this year. Typecast currently has 300 Korean voices and 100 English voices. It also offers services in Japanese and Spanish. This year, it plans to launch Chinese, German, Italian, and Portuguese to expand its user base. CEO Kim said, "Neosapience was created anticipating the popularization of AI virtual humans who can see, hear, speak, and act like people." He added, "We will create a future where AI that can naturally express emotions and interact like humans can be encountered not only in various media but anywhere in our daily lives."