Dare to build the next generation of realtime translation devices powered by Edge AI
Presented by Saigon AI Hub and Qualcomm
Designing a Future Where Language Is No Longer a Barrier
The problem
Technology exists, but communication still breaks down
In today’s global economy, language is no longer just culture; it determines access to opportunity.
Across factories, construction sites, logistics hubs, hospitals, and service environments, workers from different countries must collaborate every day. Productivity, skills transfer, workplace safety, and career mobility increasingly depend on the ability to communicate clearly and instantly across languages.
Yet communication inequality remains a hidden barrier. The reality today:

Only 5% of Vietnamese workers are proficient in English, lower than many non-English speaking ASEAN countries
ManpowerGroup, 2022
87% FDI companies require English or foreign language proficiency from employees
Phan & Dang (2024), IJMCER
Years to learn: Language fluency requires long-term training that many workers cannot afford
Gestures or mixed-language communication are often relied on, increasing misunderstanding and risk
Connectivity barriers: Most translation tools rely on stable internet or cloud processing, which conditions are not always available in real-world environments
The challenge
Designing communication without interruption
OneVoice AI Challenge calls on innovators to rethink how humans communicate across languages.
Teams are invited to design standalone AI translation devices powered entirely on-device using Edge AI.
The goal: Enable natural bilingual conversations with minimal latency, even in noisy, hands-busy, or low-connectivity environments.
Teams will design and prototype devices that combine:
Edge AI models
On-device ML combining speech recognition, translation, and synthesis - no cloud dependency, optimized for speed and accuracy.
User-centered design
Intuitive device interaction designed for workers in factories, construction sites, and logistics centers, with minimal training required.
Note: Contestants are encouraged to use models from the Qualcomm AI Hub, which contains hundreds of models (like ControlNet, Stable Diffusion, and Llama) that have already been "pre-optimized" (quantized and compiled) to run perfectly on Snapdragon platforms.
Technical Requirements
Solutions may be deployed on any portable device format, without restriction to predefined device categories.
Core specifications: Solutions will be evaluated against the following technical benchmarks:
Innovation & Novelty (25%): Evaluates the problem-solving approach and the effectiveness of supplementary features.
Technical Performance (50%): Assesses core on-device system metrics, including accuracy, duration, speed, and latency.
Business Viability (25%): Evaluates the potential for productization and commercialization.

Target languages:
Vietnamese ↔ English
Vietnamese ↔ Mandarin
Vietnamese ↔ Korean
Specific benchmark values will be published prior to the prototype submission phase. The final round will feature live demonstrations and winner announcement, where shortlisted teams present their working prototypes to a panel of judges.
Judging criteria
Solutions will be evaluated based on 3 core dimensions:
🚀 Innovation & Novelty - 25%
We celebrate ideas that challenge conventions and introduce new ways to bridge language barriers. Judges will look at how creatively teams approach the problem, the originality of the solution, and meaningful features that enhance real-world usability.
⚙️ Technical Excellence - 50%
At the heart of the challenge lies reliable Edge AI performance. Teams will be assessed on the effectiveness of their on-device translation system, including: translation accuracy, processing speed and responsiveness, low latency communication, and stability and operational performance.
🌍 Business Impact & Real-World Potential - 25%
Judges will evaluate how feasible each solution is for real deployment, including its practicality, scalability, and potential to evolve into a commercially viable product that benefits workers and organizations at scale.
Meet the Judges
Guided by experience and driven by innovation, our judges are experts from industry and academia who will challenge, evaluate, and spotlight the most impactful AI solutions.

Dr. Châu Thành Đức
Lecturer, Faculty of Information Technology, HCMUS, VNUHCM
SAIH Executive Board

Mr. Võ Trọng Thư
AI Manager, AI Lab, GreenNode
SAIH Executive Board

Prof. Dr. Quản Thành Thơ
Dean, Faculty of Computer Science & Engineering, HCMUT, VNUHCM
SAIH Executive Board

Mr. Hoàng Ngọc Thức
Product Marketing Manager
Qualcomm Vietnam

Ms. Nguyễn Thanh Thảo
Staff Manager, Business Development
Qualcomm Vietnam
Participation Journey
Who should join
OneVoice AI Challenge welcomes builders from across the global AI ecosystem, with a strong focus on innovators from Vietnam and Asia. Teams of any size are welcome to participate:
Independent researchers
AI practitioners and engineers working independently or in small teams
Startups & small teams
Early-stage startups or small product teams building practical AI solutions
University teams
Student and research teams combining academic research with applied engineering
Additional eligibility conditions
The submitted solution must be an original development created specifically for this competition.
Teams must not have an already commercialized product that is substantially similar to the competition submission.
Products that are currently on the market, revenue-generating, or in mass production are not eligible.
Early-stage prototypes or pre-commercial concepts are allowed.

Competition timeline
OneVoice AI Challenge is structured as a multi-phase innovation program, guiding teams from idea to working prototype through several development stages. The competition follows a hybrid format.
Phase 1 - Team registration
Phase 2 - Technical specification submission
Participants submit system architecture and technical design outlining their proposed solution.
Technical workshop(s) available
Phase 3 - Prototype submission
Selected teams build functional device prototypes and prepare demonstration scenarios.
Technical workshop(s) available
Phase 4 - Field testing and evaluation
Prototypes are evaluated in real-world communication scenarios to assess performance and usability.
Device support provided
Phase 5 - Finale and live demonstration in Ho Chi Minh City, Vietnam
Finalists present and demo their solutions to a panel of industry experts, researchers, and technology leaders.
Awards & Participation support
Awards & recognition
Top-performing teams will receive recognition for their technical innovation and practical impact:
Total prize pool: $50,000
First Prize
$20,000
+ Partnership for product development
Second Prize
$15,000
+ Incubation support
Third Prize
$10,000
+ Incubation support
Special Innovation Awards
Total $5,000
for breakthrough innovations
Participant support
Throughout the program, teams will gain access to valuable resources designed to accelerate development:
Technical mentorship from experts in Edge AI and speech technologies
Access to development resources and optimization guidance for Qualcomm Snapdragon platforms
Global exposure to international researchers, industry leaders, and investors
Prototype development support to help teams move from concept to fully functioning prototype
Powered by a global innovation ecosystem
Saigon AI Hub was founded in 2025 as Vietnam’s first open research space connecting researchers, experts, lecturers, and the AI community in Ho Chi Minh City. The initiative is jointly established by VNG Group and Vietnam National University Ho Chi Minh City to foster a sustainable ecosystem for applied AI research and innovation.
Latest updates
Stay informed with the latest information about the EdgeVoice Global Challenge.
Frequently asked questions
Continue the Build-to-Deploy JourneyContinue the
Build-to-Deploy Journey
Stay connected with OneVoice AI Challenge as teams transform ideas into practical Edge AI solutions


