The research work to help increase the accuracy of real-time speech recognition models (Streaming Automatic Speech Recognition) by Le Duy Khanh - "GenZ" engineer of Zalo AI - will be announced for the first time at the International Scientific Conference, taking place in Greece in September 2024.
With the topic " Improving Streaming Speech Recognition With Time-Shifted Contextual Attention And Dynamic Right Context Masking " , the research paper of the Zalo AI engineer born in 2000 achieved an almost perfect score - 11/12 points, passing the rigorous review round with more than 2,000 participating papers to be presented at the Interspeech Conference in the form of an oral session.
“ I am very proud that my first scientific article was recognized by a prestigious scientific conference and I have the opportunity to introduce Vietnam's research achievements to big-tech, experts and the international community ,” Le Duy Khanh shared.
Under the guidance of Dr. Chau Thanh Duc - Head of Research and Development Department at Zalo AI, Lecturer at University of Natural Sciences (Ho Chi Minh City National University), this research project is expected to make an important contribution to upgrading speech recognition models, increasing the accuracy of voice dictation and voice-to-text on Zalo application.
“ Synthesizing Zalo AI’s highly practical research into scientific papers and presenting them at prestigious international conferences is very meaningful. It not only demonstrates the capacity of Vietnamese engineers, but also demonstrates the desire to share experiences and contribute to the development of the global AI community,” said Dr. Chau Thanh Duc.
Previously, Zalo integrated this research into its messaging application from the end of 2023, helping to significantly improve the accuracy of the "voice message composition" feature. This feature allows users to compose messages by voice instead of typing by hand, saving time and making it more convenient in many usage situations. At the same time, the accuracy of this feature has reached 95% in practice; the rate of needing to re-edit text after composing by voice has decreased from 6.4% to only 4.8%.
According to Zalo statistics, although the feature is still in the testing phase, it has generated nearly 4.5 million messages per day and attracted about 3.2 million monthly users (data updated to June 2024).
Since starting its pioneering journey in AI research in 2017, Zalo has always believed in “empowering” the younger generation. Currently, up to 31% of Zalo employees belong to the GenZ generation. In 2021, two other research topics of the Zalo AI engineering team related to speech processing technology were also recognized at the Asia- Pacific International Conference on Artificial Intelligence (PRICAI 2021). Notably, the authors of these two topics are all young researchers under the age of 30.
Interspeech is a long-standing, comprehensive and prestigious international conference on Speech Processing organized by the International Speech Communication Association. This year, the conference with the theme “Speech and beyond ” takes place from September 1-5, 2024 on the island of Kos (Greece).
![[Photo] Binh Trieu 1 Bridge has been completed, raised by 1.1m, and will open to traffic at the end of November.](https://vphoto.vietnam.vn/thumb/1200x675/vietnam/resource/IMAGE/2025/10/2/a6549e2a3b5848a1ba76a1ded6141fae)
Comment (0)