~/Portfolio$Jude Mawad Projects

[001]Project / The Cube

THE CUBE

Voice-controlled physical computing system

A compact hardware and software system combining voice interaction, a 64×64 LED matrix, integrated audio and a custom-designed enclosure around a Raspberry Pi 5.

Exterior of The Cube showing its LED matrix and speaker grille
Product Render / Loading Interactive object / Drag to rotate

[002]

What Cube can do

What Cube can do

  1. Movies through my server

    I can ask Cube to download a movie through my home server, clarify which title I mean, and check its progress later. It announces when the download starts and when it’s ready. I can also cancel a tracked request after confirmation.

    Explore my server’s movie automation
  2. Music on either speaker

    I can play tracks, artists, and playlists through Cube’s built-in speaker or a connected Bluetooth speaker, and switch between them by voice. My phone controls the same Spotify session. Music and assistant speech have separate volumes, and music gets quieter while I talk to Cube.

  3. Local AI with tools

    For conversations and more flexible requests, Cube connects to an LLM running on my PC. Alongside answering questions, it can request actions for movies, lighting, plug status, and Cube’s controls. The backend checks and executes those requests. The PC also provides speech recognition and synthesis.

  4. An animated display

    Between interactions, the 64×64 display cycles through a falling-block clock, animated weather, pixel animations, and Spotify artwork with playback progress. It also shows when Cube is listening, thinking, or speaking.

  5. Lights and plugs

    I connected Cube to my Govee lights and Tuya devices so I can control them without opening their apps. I can switch individual devices or groups, adjust supported light brightness and colours, and check whether a plug is on.

  6. Follow-ups and interruptions

    When Cube asks for clarification, I can answer without repeating “Hey Cube” during the follow-up window. I can also interrupt a spoken reply with another wake. Movie requests support corrections such as “the newer one” or “no, I meant…”.

My PC doesn’t have to stay on. Supported music, lighting, plug, and device commands work without the AI node, using speech fallbacks on the server and Pi. The backend still needs to be running.

[003]

Display apps

Display apps

Rendered previews from Cube’s display apps, using sample data.

Cube with its LED display and speaker
  1. Clock

    I went with a Tetris-style clock instead of plain digits. Falling blocks build the time, with the date sitting underneath.

  2. Weather

    I kept the weather screen simple: the current temperature, chance of rain, and a small animated scene that changes with the conditions.

  3. Animation

    I added this mostly for fun. I can convert GIFs for the 64×64 panel and have them rotate between the other screens.

  4. Music / Spotify

    I connected Cube to Spotify so I can control music by voice or from my phone. The display shows the album cover and track progress.

[004]

Physical Design

Physical Design

Explore the structure, component layout and enclosure of the Cube.

05 / Hardware

WHAT'S INSIDE.

These are the main components inside the Cube.

  • Raspberry Pi 5

    Compute · 8 GB

  • 64×64 LED Matrix

    RGB display

  • ReSpeaker XVF3800

    Microphone array

  • Visaton FR 7/4

    Full-range speaker

  • HUSB238

    USB-C PD negotiation

  • 52Pi power board

    Pi power delivery

  • Pololu 5 V / 9 A regulator

    Voltage regulation

  • Noctua fan

    40 mm · Active airflow