Entering the airport lobby, amidst the hustle and bustle of people returning from a business trip, Ho Minh Duc paused for a few seconds when he heard a familiar, gentle female voice reading an announcement on the system.
Vbee's staff are working at the company headquarters in Hanoi - Photo: NVCC
He smiled, feeling relieved and happy as if he had met a relative. That "relative" was one of the 20 AI voices that Duc and the Vbee team had "eaten and slept" with for many days and months, pouring all their heart into every sound line, taking care of every nuance of the voice to make them more natural and human-like.
Bumpy start-up
I don't know how many times CEO Ho Minh Duc and CTO Nguyen Thi Thu Trang - the two founders of Vbee Data Solutions and Services Joint Stock Company - have experienced such a feeling of joy and pride.
They met "special acquaintances" in many different situations: clear voices on school loudspeaker systems, warm voices in buildings, or professional voices from the automatic switchboards of many businesses.
Vbee's brainchildren are no longer just the result of algorithms and codes, but are actually entering life, making silent but powerful contributions to many fields.
From book introductions, movie dubbing to automated call center announcements, Vbee breathes new life into voice technology.
As the "mother" of the core TTS technology, Dr. Nguyen Thi Thu Trang always aspires to bring products from Vietnamese speech synthesis technology - the technology that she has devoted a lot of effort to since her doctoral thesis at Paris 11 University - to real users.
Vbee’s early days were rocky. Despite being free for the first two years, its text-to-speech (TTS) tool attracted only a small audience. But then COVID-19 turned into an unexpected turning point.
Faced with strict social distancing regulations, businesses like FE Credit, Momo, Viet Credit, Sacombank... had to find ways to reach thousands of customers. That's when Vbee was given the opportunity: from debt reminders to automatic responses, their products promptly became the optimal solution. At that time, virtual assistants and virtual call centers brought in up to 80% of Vbee's revenue.
When the pandemic passed and the world economy went down, Vbee faced a new challenge. The wave of generative AI (GenAI) and digital content trends revived the TTS tool. Today, from TikTok to YouTube, Facebook, Vbee's AI voices are everywhere.
"A lot of TTS content is currently provided by us," Mr. Ho Minh Duc proudly shared. Currently, the number of actual users of Vbee has exceeded 2 million, and this number is still increasing steadily by 20% every month.
Vbee has trained over 20 high-quality corporate voices, and if you count custom voices, they have created over 200 different AI voices.
With the new voice transcription technology that was recently researched and tested, a new voice now only needs 3 minutes of recorded data to train instead of 4 to dozens of hours of recording like two years ago.
CEO Ho Minh Duc and Chief Technology Officer Nguyen Thi Thu Trang - two founders of Vbee Data Solutions and Services Joint Stock Company - Photo: NVCC
"We are better at understanding Vietnamese"
In the race for speech synthesis technology, CEO Ho Minh Duc sees a time when technological innovation efforts will gradually reach their limits.
According to him, Vbee is not only developing core technology for processing Vietnamese speech, but has also been building a technology system capable of deeply understanding the Vietnamese language - with all the subtleties, tones and unique culture that only true Vietnamese people can fully understand.
As the leading company in the TTS market in Vietnam, the two leaders of Vbee believe that their tool has become the standard for AI voice reading for Vietnamese. Users not only appreciate the accuracy but also feel the "emotion" in each voice developed by Vbee.
In Vietnamese, for example, just the word "alley" has many different names depending on the region such as "hèm", "kiệt", "xếc" - each word has a different nuance that AI needs to understand.
To achieve that, Vbee has invested heavily in collecting sample data sets as well as investing in powerful server systems for AI training.
"To help AI understand and process each regional nuance correctly, we had to build countless sample sets, and the cost of the processing server was also very high," CEO Ho Minh Duc shared.
Dr. Nguyen Thi Thu Trang has spent more than 15 years researching Vbee's core TTS technology to decode the unique tones and grammar of Vietnamese. For her, her mother tongue is a subtle world full of expressive nuances.
"My Vietnamese language is very complex and interesting, the tones are the most difficult and different from many other popular languages in the world. The more I understand the language, the more accurate my model will be," she explained.
Vbee is gradually asserting that they will be an indispensable part of tools and devices with integrated Vietnamese language processing software in the technology era.
In every word, every voice, the Vbee team not only researches and develops technology but also strives to create a truly "Vietnamese emotion" in their AI voices.
The name Vbee is an abbreviation of the phrase "Vietnamese BE your Eyes", which comes from my initial desire to build a tool that becomes the "eyes" for the visually impaired. But in the current development trend, when many people want to switch to listening more than seeing, we believe that Vbee will also become the "eyes" of everyone.
Dr. Nguyen Thi Thu Trang (Lecturer, School of Information Technology, Hanoi University of Science and Technology, Founder and Technology Director of Vbee Company)
Meeting of audiobook lovers
Vbee was born from the relationship between Dr. Nguyen Thi Thu Trang and the blind community. Since her student days, she has participated in recording audiobooks and developing a Vietnamese reader to support the blind.
These experiences inspired her to develop Vietnamese reading software - the predecessor of Vbee. In 2018, she and Mr. Ho Minh Duc - a classmate at Hanoi University of Science and Technology with experience from the Socbay.com project and digitizing audiobooks - founded Vbee, a pioneer in the field of text-to-speech conversion in Vietnam.
Vbee's Outstanding Achievements
- First prize of Qualcomm Vietnam Innovation Challenge 2024
- Special Prize Tuoi Tre Start-up Award 2023
- Winning start-up in Grab Venture Ignite 2020 Accelerator program
- First prize of Vietnamese Talent 2018, second prize of Vietnamese Talent 2020
- Certificate of Vietnamese Core Technology in the National Digital Transformation Program 2025 - 2030 of the Ministry of Information and Communications
- Winning project in Vietnam Digital Media Award 2018 and Vingroup Fund 2019.
Regional vision
After affirming its position in the Vietnamese market, Vbee is aiming to expand to Southeast Asia with plans to bring its TTS technology to countries such as Laos, Thailand, Cambodia and the Philippines by 2026.
According to Dr. Nguyen Thi Thu Trang, the rapid advancement of technology today with the emergence of multilingual models will make it easier to develop TTS tools for other languages.
Currently, she is researching speech technologies for Thai, Chinese and English, opening new steps for Vbee in the international market.
Vietnamese Start-up Honored at AI Summit ParisSource: https://tuoitre.vn/vbee-va-no-luc-chap-canh-cho-tieng-viet-20250217102146767.htm
Comment (0)