KittyGatto — Bilingual Learning Device for Children
A standalone, cat-shaped bilingual learning device designed for children aged 2–8. Features an E-ink display, embedded dual-language audio, physical controls, and an optional on-device AI speech recognition engine optimised for children's voices.
The Challenge
Bilingual families and early-years educators face a practical problem: good language learning tools for young children (aged 2–8) don’t exist as standalone, screen-free devices. Tablets are too general-purpose, too distracting, and unsuitable for very young children. Apps require parental mediation. The gap was a purpose-built device that a toddler could use independently, that parents could trust, and that delivered genuine bilingual learning value.
KittyGatto needed to be physically appealing (the cat form factor is deliberate), durable, simple to operate with physical buttons, and technically capable — with an E-ink display for eye comfort, embedded audio, and an AI speech recognition layer that could actually understand young children’s voices.
The Approach
We led the hardware and embedded systems development for KittyGatto, working across the full stack from electronics design to firmware and the edge AI layer. Key work included:
- Hardware design for the cat-shaped enclosure, balancing aesthetics, durability, and electronics accessibility, working in partnership with Barclays Eagle Labs facilities
- E-ink display integration — selecting and driving the right panel for low-power, child-friendly content presentation
- Embedded firmware for the dual-language audio system and physical button interface
- BLE connectivity for content updates and parental controls without requiring the device to be always-connected
- Low-power architecture to maximise battery life for a device used by children
- Edge AI speech recognition development, training and optimising a model for children’s voices specifically, which differ significantly from adult speech in pitch, disfluency, and pronunciation
- End-to-end testing
The edge AI component was technically the most demanding: children’s speech recognition is a hard problem that most general-purpose speech models handle poorly, and running it on-device (without cloud dependency) required careful model selection, testing and optimisation.
The Outcome
- Working hardware product designed and prototyped
- E-ink display with dual-language audio operational on physical device
- Physical buttons interface child-tested and refined
- BLE content update system operational
- Pro version with edge AI speech recognition developed for children’s voices
Technologies & Methods
Embedded C firmware, E-ink display integration, I2S audio, BLE (Bluetooth Low Energy), low-power MCU design, edge AI (on-device speech recognition, voice model optimisation for children), hardware prototyping, DFM review.
About this engagement
This was a Build engagement: creating a hardware-software consumer product from concept through to launch. The combination of embedded systems, edge AI, and child-centred design represents some of the most technically demanding product development work — every constraint is tight and the end user is a young child.