🎯 Inspiration Communication should be seamless and inclusive, but language barriers often make it challenging. Whether in business meetings, classrooms, or everyday conversations, the need for real-time speech translation is more crucial than ever. Instead of relying on WiFi, why can't it be an offline device that can be used by children and the elderly just by pressing a button? Inspired by this, we created VerbaLink—a wearable device that transcribes and translates speech instantly, making conversations accessible to everyone.
🛠 What It Does VerbaLink is a real-time speech transcription and translation system that: ✅ Records spoken language using an embedded microphone. ✅ Transcribes audio with high accuracy using Faster Whisper AI. ✅ Translates speech into multiple languages via Google Translate. ✅ Transfers files & text over Serial to integrate with external systems.
It’s fast, portable, and efficient, making it perfect for education, accessibility, international communication, and more.
🔧 How We Built It We combined hardware and software for a seamless experience:
Arduino + TMRpcm: Handles real-time audio recording and playback. SD Card Storage: Saves WAV files for processing and transcription. Faster Whisper Model: Provides highly accurate speech-to-text transcription. Google Translate API: Converts text into multiple languages. Python Serial Communication: Ensures smooth data transfer between hardware and software.
🚧 Challenges We Ran Into 🔴 SPI & I2C Conflicts: The design was supposed to display the received text on an LCD, but there was either a library conflict or a memory overload. 🔴 Hardware Limitations: The entire transcription process was supposed to be done by the Arduino itself, but the computing power is not enough. 🔴 Serial Data Corruption: WAV file transfers were sometimes corrupt due to buffer overflows. 🔴 Real-Time Optimization: Ensuring low latency transcription & translation while maintaining accuracy was tricky.
🏆 Accomplishments That We're Proud Of 🌟 Built a fully functional speech transcription & translation device. 🌟 Resolved complex hardware-software integration issues. 🌟 Achieved real-time processing with minimal latency. 🌟 Created a system that can be expanded for accessibility & multilingual applications.
📚 What We Learned 💡 Efficient resource management is key in embedded systems. 💡 Recording audio signal with a mic amplifier 💡 Soldering is not as easy as YouTube tutorials 💡 SD card integration in Arduino systems. 💡 Python-Arduino serial communication requires buffering & proper timing for large file transfers. 💡 Transcription models like Faster Whisper can achieve high accuracy with the right parameters.
🚀 What’s Next for VerbaLink? 🔹 Chip improvement: Use a better chip to transcribe the data without transferring the data to a PC 🔹 Voice Output (TTS): Convert translated text into natural speech playback. 🔹 Multi-Speaker Support: Differentiate voices for group conversations. Able to translate from English to other languages by adjusting the config on a menu. 🔹 Smaller & More Portable Version: Optimizing for wearable devices or smart glasses.
VerbaLink is just the beginning. The future of real-time multilingual communication is here!
Log in or sign up for Devpost to join the conversation.