AsiaTechDaily – Asia's Leading Tech and Startup Media Platform
AI-based Voice Technology is changing the ways the world communicates. The technology is used in several spaces like medical, art, e-commerce, games, Virtual Reality, etc. When used in the medical field, AI-voice technology gives accurate diagnosis, results and prevents any human led errors. Korean startup DOUB took all these factors to design its recognition service SpeechEMR, an automatic voice recognition service that records medical events and converts them into text data in real-time facilitating users to record medical events instantly.
Technology to aid precise Medical diagnosis
SpeechEMR provides a high recognition rate of over 95 percent using Artificial Intelligence (AI) voice recognition technology specially designed for use in the medical field. Spoken audio data such as the conversations between doctors and patients or medical dictations are converted into text in real-time, through processes such as noise removal and silent syllable separation.
This voice recognition service then quickly edits and saves the medical records with misspelling and omission on display coupled with correct word suggestions and medical terminology dictionary provision. Preventing information errors and improving the clarity highlights important information such as numbers, dates, units, sizes, and locations, increasing the clarity and preventing sensitive information errors.
DOUB’s technology used in various sectors
The Seoul-based startup DOUB is a data analytics company that was established in 2015. . In April 2018, it launched a text analysis service that combines deep learning and search algorithms. Within a year, it supplied products to six contact centers of major financial institutions. It also successfully commercialized the industry’s first real-time text analysis service, which translates conversational voice into text, evaluates compliance, and immediately recommends knowledge related to counseling topics.
In 2019 at the POSCO Idea MarketPlace, a Korean startup competition, DOUB was among the top ten finalists dubbed as an investment target as a result of which it attracted seed investment. In 2020, the startup won the AI award at the Hi-Tech Award function.
“DOUB has been growing at a pace of more than 30 percent annually after it attracted seed investment in 2019, targeting the corporate market,” said its founder Son Hyeon-kon. “We are now preparing to invest in Series A to expand our business, and through this, we will enter the B2C market so that anyone can use real-time text analysis technology to lead the popularization of AI services,” he added.
Currently, DOUB’s solutions are being utilized in the various financial sector. Many large contact centers like Mirae Asset Securities and Samsung Fire and Marine Insurance, Korea’s No. 1 insurance companies, are using DOUB’s text analysis solution.
The first customer of DOUB’s medical conversation analysis solution is Hanyang University Hospital and their main area, the conversation analysis market, is targeting the market in cooperation with contact center system integrators such as Buttle Information System and Hankook Cloud Inc.
SpeechEMR guarantees a high recognition rate by optimized adaptive training through AI speech recognition technology, medical-specific acoustic or language models, and terminology dictionary. SpeechEMR’s language correction feature uses 500,000 words of everyday terms and 100,000 words from medical terminology.
DOUB is currently developing a real-time conversation analysis service platform that integrates its natural language processing-based AI technology that has been accumulated over the years. The service will allow individuals with small businesses such as private operators, small shop-owners, schools, and small hospitals to utilize text analysis services at a low cost through the cloud. The platform will equip its users with various conversational analysis services such as conference records, virtual secretaries, call summaries, counseling robots, counseling notes, and marketing analysis.