Van Dinh Hong Vu co-founded Elsa Speak, an AI-powered English learning app that has helped tens of millions of users from over 100 countries improve their English pronunciation.
At the World Economic Forum (WEF) in Davos in late January, Hong Vu was among 200 founders recognized as Technology Pioneers. The Elsa (English Language Speech Assistant) app, which she and her team have developed over nearly a decade, has become one of the most popular English learning applications globally, reaching more than 50 million users.
The app was recently featured by Apple on the App Store in Vietnam for the 2025 New Year, highlighting it as a prime example of how technology, especially AI, can transform education. "Elsa Speak uses a custom-built AI model trained to analyze users' pronunciation at the syllable level, providing detailed feedback on possible improvements," according to the App Store's review of the product.
Believing in AI's Power
The success of Elsa Speak shows how artificial intelligence can make language learning more accessible and effective for people worldwide. The app's innovative approach to pronunciation training has made it a valuable tool for anyone looking to improve their English speaking skills.
Despite having strong English skills during her high school years before studying abroad, Vu admits she still faced communication challenges in her early days overseas. The Stanford University graduate believes this common problem stems from limited access to native-speaking teachers in her home country, making it difficult for students to practice proper pronunciation. She also points out the challenge of finding one-on-one learning opportunities where learners can receive immediate feedback on their mistakes.
"Technology is the only solution to overcome this challenge," she explains.
By 2025, AI has become a buzzword appearing in almost every software product. However, when the Elsa project started in 2015, the concept wasn't as widespread, mainly existing in virtual assistant applications. Initially, the team considered using Google's speech recognition API for their project. However, this technology, designed for virtual assistants, allowed for significant error margins - meaning it would understand and execute commands even when words were pronounced incorrectly. This feature, while useful for virtual assistants, would actually hinder language learning progress since learners wouldn't receive accurate feedback on their pronunciation mistakes.
She collaborated with Dr. Xavier Anguera, an expert in speech recognition and AI, to develop a unique solution that uses deep learning and artificial intelligence to detect pronunciation errors. The results were impressive, with an estimated accuracy rate of over 95%. Even in a long conversation, the system could identify minor stress or intonation mistakes in individual words. Using AI, they also developed a Speech Analyzer feature that provides users with detailed feedback and suggestions for improving their speaking skills during meetings, online calls, and other real-world situations.
When generative AI began to gain momentum in late 2022, ELSA was among the pioneering projects to implement new solutions in communication practice through its ELSA AI feature. "This feature allows users to engage in simulated conversations tailored to their personal learning goals," Vu explains. "It helps them practice English communication in contexts that matter most to them, whether for work or daily life. The AI-powered conversations feel natural and provide immediate feedback, making it easier for learners to improve their speaking skills in a comfortable environment. Users can practice specific scenarios they might encounter, building confidence before facing similar situations in real life."
Users can engage in natural, everyday conversations with their "AI tutor" on any topic without following rigid patterns. If they run out of ideas or don't know what to say next, the AI provides suggestions in both English and their native language, even offering written prompts to enhance their communication abilities.
"When practicing topic-based conversations, learners can apply practical grammar and vocabulary. With instant feedback and suggestions for improving grammar, vocabulary, tone, pronunciation, and fluency, their English proficiency naturally develops," explains Vu.
According to ELSA's founder, smartphones now enable one-on-one learning experiences. This innovative approach addresses the challenge of teacher shortages by ensuring each learner receives support from an "AI teacher" that adapts to their real-world needs. Students benefit from a more personalized learning path and can study anytime, anywhere.
From Vietnam to the World.
This approach makes language learning more accessible and effective for international students. The AI-powered system provides a supportive environment where learners can practice at their own pace, making mistakes without fear and receiving constructive feedback to improve their English skills continuously.
Although ELSA started in the United States, most of its new features were first introduced to Vietnamese users. According to CEO Van Dinh Hong Vu, during the initial phase, 80-90% of app users were from Vietnam. Today, while the number of domestic users continues to grow and makes up the majority of registered accounts, their proportion has decreased to around 20%, indicating the app's successful expansion into many other markets.
Having experienced communication challenges during her first year studying abroad, Vu noticed that many of her international friends faced similar language barriers. This observation highlighted a significant potential market for startups focusing on conversational English learning. According to ELSA's website, the app has now reached an impressive milestone, being downloaded and used in 195 countries and territories, with over 50 million users as of 2024.
After being recognized by Forbes as one of four companies using AI to change the world in 2017, Vu notes that users' attitudes toward AI have evolved significantly. People are gradually accepting this technology as an essential part of their work life, helping them save time for higher-value tasks such as creative thinking and strategic planning.
"This aligns perfectly with ELSA's vision to develop an AI-powered application that helps users learn more effectively, saving both time and money so they can focus on what truly matters," she explains.
With the rapid advancement of generative AI, Vu notes that English learning applications need to evolve and adapt more quickly. The startup aims to transform into an AI communication platform that enables users to converse confidently in any situation with anyone. Their goal is to help learners think and react naturally in English, ultimately achieving fluency in their conversations.
"Each user will follow a personalized learning path and content structure that's more specialized, allowing them to improve their English skills while developing other valuable competencies simultaneously," she adds. This approach reflects ELSA's commitment to providing a comprehensive learning experience that goes beyond traditional language instruction.
-----
Source: https://vnexpress.net/co-gai-viet-dung-ai-day-tieng-anh-cho-hang-trieu-nguoi-4843380.html
0 Comments