The AI technology sector in Korea is on the rise. AI is now incorporated into smartphones via virtual assistants such as in the case of the iPhone’s Siri. South Korea has its own “Siri” and it is Clova, an AI virtual assistant developed by Korean Internet giant Naver. Naver is the top search engine in South Korea, not Google. Most global IT companies have their own voice recognition system but they are in English which makes sense since they are looking to make their company a global company.
Naver took their time to develop CLOVA. The reason for this was the fact that there wasn’t enough data collected to create a Korean voice recognition system until recently. The Korean language is not as complex as many would believe.
Even IBM’s Watson has been learning/studying the Korean language for many years now, however, it still has some issues. Many who have used IBM’s Watson feel their Korean proficiency is at a teenager’s level. For a computer to completely master a language is difficult.
Speaking a language, the way it is spoken in real life is a very tricky thing as it is not something you can learn from a book or have programmed. Therefore, the only system that has the potential to master the Korean language fully is a system that is made by the operator of Korea’s largest search engine, Naver. Naver has with them an enormous amount of data in Korea. Currently, it offers translation services and some voice recognition services in Korean. Furthermore, Naver was able to team up with local language institutions to further advance the ability of its virtual assistant.
What is Clova by Naver?
Naver launched CLOVA back in 2017. CLOVA is a voice assistant, especially for the Korean language. The data analysis and artificial intelligence technology were based on the same system used by Apple. Users can use CLOVA on their smart devices like smartphones, tablets, and even CLOVA-featured AI speakers (WAVE). In addition, Naver created CLOVA CSR a voice recognition API service that allows developers to create services like chatbots, voice memos, and virtual assistants using CLOVA. CLOVA is not just for Korean voice recognition, CSR also supports English, Chinese, and Japanese.
Companies in Korea have been incorporating CLOVA into many of their existing services. Thanks to CLOVA’s ability to convert text to speech in a natural Korean voice, companies no longer need to hire voice actors. CLOVA is able to produce the Korean language in a natural way which can be used to convert news to audio and books to audiobooks. Korean company Woongjin Book Club has been able to incorporate CLOVA into their solutions to create an interactive book service that reads books to children as well as converses with them. In addition, CLOVA continues to improve since it is based on machine learning and is constantly improving itself using the data collected from Naver.
HyperCLOVA
In mid-2021, Naver launched its AI platform that can process large amounts of data called HyperCLOVA. It is Korea’s most advanced AI that is trained in the Korean language. It can analyze large datasets in order to create new AI models. For example, it can be used to create AI tools that can chat with humans and even summarize complicated documents in seconds. It is estimated that HyperCLOVA’s AI language tool has over 200 billion parameters. Compare this to Open AI which has 175 billion.
HyperCLOVA was used to build a new search engine service called Knowledge Interactive. It has the ability to not simply answer questions but engage with the user. Knowledge Interactive analyzes the intention of the user’s question and finds the best possible answer. Then it provides supporting evidence to allow users to refer to them so they can continue to explore. In addition, they are incorporating AR (Augmented Reality) technology so users can interact with 3D images. Their aim is to bring a more visual experience to current search engine experiences.
CLOVA Lamp
Last year, Naver came out with CLOVA Lamp, a smart reading lamp that helps children by reading their books to them. The CLOVA lamp incorporates machine learning and computer vision technology to convert text and images from a children’s book into speech. How does the process work?
- Image recognition technology captures the data (Image and text of the children’s book)
- The system’s cloud-based AI analyzes the data
- Convert the data into speech
There are two ways to activate the CLOVA Lamp.
- Push the button labeled “read”
- Use a simple voice command
In addition, CLOVA can explain the meaning of a particular word if a child requests it. If there are any questions a child might have, CLOVA will answer them. Therefore, this is great for parents who do not have the time to read to their children. Children need to hear words frequently to build up their imagination, critical thinking, and concentration, and improve their vocabulary.
Now children will be able to hear their favorite stories anytime they want. This will create their interest in books and build their passion for reading. The smart lamp has a list of completed books stored on the device. In addition, CLOVA can recommend different books based on their child’s interests. The lamp itself has a ring-shaped responsive LED light and a small camera built into the head.
