Soyle: Smart Glasses for Real-Time Speech Transcription
Role: Founder & Lead Developer
What started as a crazy idea in early 2024 turned into an incredible journey of innovation, persistence, and unexpected success. Soyle is a pair of smart glasses that transcribe speech to text and display it directly on the lenses in real-time – because apparently regular subtitles weren't challenging enough.
The Journey
From Idea to MVP
The concept was ambitious: create wearable technology that could help people with hearing impairments or language barriers by displaying real-time transcriptions directly in their field of vision. Across 5 hardware/software iterations, we refined the design, integrating a Raspberry Pi 4 for on-device processing and an ESP32 microcontroller for display control. By February, we had our MVP ready for Digital Almaty 2025.
Digital Almaty 2025: The Underdog Story
Plot twist: we couldn't afford the fancy 1,000,000 tenge exhibition zone, so we did what any broke but determined startup would do – set up shop in the middle of the hall and pitched to anyone who'd listen.
Best decision ever.
Erzat Dulat from Higgsfield and Ali Hasanov from Silkway Accelerator (Google Startups) stopped by, and Erzat dropped this gem: "This is one of the few projects that truly impressed me and provided something refreshing here at Digital Almaty 2025." (Never expected that reaction)
Going Viral
Then Rashit Ilyasov featured us in his reels, and suddenly we're viral – 1.7M views across TikTok and Instagram. From clunky prototype to internet sensation in 48 hours.
The attention opened doors we didn't even know existed. Meetings with Mansuri, newfound investor interest, and the motivation to go all-in.
Samsung Solve For Tomorrow 2025
We applied to Samsung Solve For Tomorrow in April, made it through semi-finals, and somehow landed in the finals.
Cue the most intense summer of our lives. While doing my internship at ISSAI NU, we were pulling late nights in their labs (shoutout to my mentor Anuar Aryngazin for letting us turn his research space into our startup garage – couldn't have done it without you).
August 21st: Finals day. Result? 3rd place and a 1,000,000 tenge (~$2,200) grant to keep pushing forward.
Technology
Core Features
Real-Time Speech Transcription: Advanced speech recognition technology processes audio in real-time and converts it to text.
AR Display Integration: Text is projected directly onto the lenses using augmented reality technology, allowing users to read transcriptions without looking away.
Low Latency Processing: Optimized for minimal delay between speech and text display, ensuring natural conversation flow.
Multi-Language Support: Designed to support multiple languages, making it useful for language learners and international communication.
Bi-Directional Communication
We also prototyped a bi-directional communication concept: a wearable camera necklace for sign language capture. This system included a baseline recognition pipeline converting signs to text and then to speech (PoC), documenting key failure modes like lighting, occlusions, and motion blur.
Hardware Components
- Microphone Array: High-quality microphones for clear audio capture
- Processing Unit: Raspberry Pi 4 for real-time transcription
- Display System: ESP32 microcontroller controlling AR lenses (OLED)
- Battery System: Long-lasting power for extended use
- Connectivity: Bluetooth and WiFi for updates and cloud processing when needed
Impact and Recognition
Awards and Achievements
- 3rd Place - Samsung Solve For Tomorrow 2025
- 1,000,000 tenge prize money from Samsung
- 1.7M views across TikTok and Instagram
- Featured by Rashit Ilyasov in viral social media content
Industry Recognition
- Praised by Erzat Dulat (Higgsfield) as one of the most impressive projects at Digital Almaty 2025
- Interest from Silkway Accelerator (Google Startups)
- Meetings with potential investors including Mansuri
The Team and Development
Development Timeline
- Early 2024: Initial concept and first prototype
- February 2025: MVP ready for Digital Almaty 2025
- April 2025: Applied to Samsung Solve For Tomorrow
- Summer 2025: Intensive development at ISSAI NU labs
- August 21, 2025: Finals and 3rd place win
Key Collaborations
- ISSAI NU: Provided lab space and mentorship during development
- Anuar Aryngazin: Mentor who supported the project during internship
- Digital Almaty 2025: Platform for initial exposure and networking
Use Cases
Accessibility
- Hearing Impairments: Real-time captions for conversations
- Language Barriers: Translation and transcription for multilingual communication
- Learning Disabilities: Visual support for auditory processing
Professional Applications
- Meetings: Real-time transcription of business meetings
- Education: Lecture transcription for students
- Healthcare: Doctor-patient communication support
- Customer Service: Enhanced communication in service industries
Clinical Testing
To validate our technology in real-world medical settings, we conducted testing at Mediker Hospital. The approbation process involved testing the glasses with patients and medical staff to ensure accuracy and usability in clinical environments.
Technical Challenges
Prototype Iterations
Five prototypes were developed, each addressing different challenges:
- Audio Quality: Improving microphone sensitivity and noise cancellation
- Display Clarity: Optimizing text visibility and readability on lenses
- Processing Speed: Reducing latency for real-time performance
- Battery Life: Extending usage time while maintaining performance
- Form Factor: Making the glasses comfortable and stylish
Solutions
- Advanced signal processing algorithms for noise reduction
- Efficient AR rendering techniques for text display
- Optimized machine learning models for faster transcription
- Power management systems for extended battery life
- Ergonomic design iterations for comfort and aesthetics
Future Development
The journey continues. With the recognition and funding from Samsung Solve For Tomorrow, we're excited to:
- Refine the Hardware: Improve form factor, battery life, and display quality
- Expand Language Support: Add more languages and dialects
- Enhance Accuracy: Improve transcription accuracy through better models
- Commercial Development: Move towards production-ready devices
- Accessibility Focus: Ensure the technology is accessible to those who need it most
Lessons Learned
Sometimes the best projects start with the craziest ideas and a refusal to give up when you can't afford the premium booth space. The Soyle journey taught us:
- Persistence Pays Off: Five prototypes and countless challenges couldn't stop us
- Networking Matters: Setting up in the middle of the hall led to our biggest break
- Viral Moments: Social media can amplify your message in unexpected ways
- Mentorship is Key: Support from mentors like Anuar Aryngazin was invaluable
- Competitions Open Doors: Samsung Solve For Tomorrow provided funding and validation
Recognition
- Featured: Digital Almaty 2025
- Winner: Samsung Solve For Tomorrow 2025 (3rd Place)
- Viral: 1.7M views on TikTok and Instagram
- Industry Praise: Recognized by Higgsfield and Google Startups
Soyle represents what's possible when you combine ambitious ideas with determination, creativity, and the willingness to pitch from the middle of the hall. From a crazy concept to viral sensation to award winner – sometimes the best path forward is the one you create yourself.
Excited to share what's next for Soyle.