EdtechEmbedded Software

Vika Cube AI Learning Companion

Embedded voice companion and a 2.4-inch interface for a children's learning device. Team of 3 over 6 months.

Key Details

Challenge	An HK-based technology company needed embedded software, voice interaction and a small display for a children's AI learning companion.
Solution	Embedded software with AI voice, a 2.4-inch TFT interface, hardware integration and companion flows.
Technologies	C/C++, Embedded Linux, RTOS, OpenAI GPT, LLM, Whisper, Speech-to-Text, Text-to-Speech, Qt/QML, Python, REST, WebSocket, MQTT, Wi-Fi, Bluetooth

Technologies used

C/C++Embedded LinuxRTOSOpenAI GPTWhisperSpeech-to-TextText-to-SpeechQt/QMLPythonMQTT

Client

The client is a Hong Kong-based technology company. Vika Cube is a compact learning companion for children, combining voice, a 2.4-inch display and flows for learning, storytelling, games and exploration. The client company is not named.

Product screenshots

Soft sky-blue and lemon scene with a rounded cube-shaped learning device whose small screen shows an animated face and audio waveform, LED ring and buttons, speech bubbles between a child and the device about why the sky is blue, and a strip of four small screen states labelled idle, listening, thinking and speaking or quiz. — The device: a child asks the cube a question, and the 2.4-inch screen animates through listening, thinking and speaking while the voice reply plays.

Dark technical architecture diagram split into an on-device firmware block with display, button, LED, microphone, speaker and wake-word modules and a cloud block with speech-to-text, language model, text-to-speech, content and OTA update services, connected through Wi-Fi and an MQTT broker, with a latency breakdown and a serial console log underneath. — Inside the device: firmware modules for the screen, buttons, LEDs, mic and speaker talk to the cloud voice pipeline over MQTT, with a live serial log of one conversation turn.

What we built

Embedded software for the display, buttons, speaker, microphone, LEDs and core device functions
Voice interaction so a child can speak to the companion and hear a reply
2.4-inch TFT interface with animations during interaction
Hardware integration, including microphone, speaker, buttons, status LEDs, display, USB-C charging and device controls
Companion flows for learning, storytelling, games, exploration and emotional companionship
Software tuned for a compact, low-power device

Engagement

Delivery was time and materials with 3 engineers over 6 months.

Project team: 3 engineers. Duration: 6 months. Model: Time & materials.

Services used

AI Services →Embedded Software →Voice AI Agent →