J a m b o x x   I n f i n i t e

Discovery & Planning: In October we officially launched the project with a Kick-off Meeting with the client, where we discussed goals and key success indicators. We then proceeded with detailed requirement clarification and conducted user research with school children and teachers to understand their musical habits and expectations.

Based on the gathered insights, we finalized the core features and held our first client review session. In parallel, we began preparing an HCI report to document the user interaction aspects. For visualization, we created personas (Persona Photo 1 and Photo 2) representing our future users.

Persona Photo 1:

Persona Photo 2:

Project Setup / Infrastructure: In November, we submitted the HCI report and proceeded with setting up the project’s infrastructure. Our Git repository (GIT Picture) played a key role as we configured basic branches, CI/CD, and coding standards. We also conducted a technical feasibility review and held a second client review session.

Git Picture:

In parallel, work on artificial intelligence began: we started setting up the backend for AI Voice Cloning and the AI Music Teacher. On the frontend, we chose Unity and established the initial multiplayer network setup. Additionally, we launched development of virtual instruments and began designing the first instrument – the drums (Drums photo). Following a further client review, we started developing the piano and ukulele instruments to build a comprehensive musical environment.

Drums Photo:

Prototyping & System Setup: In December we completed the development of virtual instruments, adding motion input control (Piano face navigation photo). We also finalized the drums and fully implemented updates for the piano and ukulele instruments. The client provided further feedback, which we incorporated during the final refinement of design and logic.

Piano face navigation photo:

Additionally, we initiated and quickly completed the development of a fourth instrument – the harmonica. At the same time, we began work on the backend logic for AI Voice Cloning to enable cloning of a singer’s voice so that the cloned voice could be used as a backing vocal.

Integration & Further Development: In January we continued refining the backend logic for AI Voice Cloning, striving to improve synthesis quality and stability. We also held a major client review session – our first official presentation to clients and partners. We received many positive comments along with constructive criticism, and subsequently focused on the most critical suggestions.

Concurrently, we started integrating minigames (Minigames photo) designed to make the learning process more engaging (Minigames photoO), especially given that our target audience includes children.

Game 1 photo:

We developed prototypes for two minigames (Minigames 1 and Minigames 2) and tested them (Game photo). At the end of the month, we began developing real-time multiplayer networking logic so users could experience playing together as a real musical group (Multiplayer Real-time Networking Logic).

Game 1 photo:

Testing & QA: In February we held two additional client review sessions. The main focus was on completing the implementation of real-time multiplayer networking and refining the backend logic for AI Voice Cloning (Voice photo). The voice cloning feature underwent significant changes and now allows users to change the singer’s voice in any song to a selected trained model.

Voice photo:

Additionally, we rapidly developed a third minigame — after deciding to cancel a fourth minigame following a client call—and implemented the backend logic for the AI Music Teacher, which helps users generate lyrics and answers music-related questions. In parallel, we created a settings page and began full-scale integration of the frontend design. The result was a unified application style (Dashboard photo) where all functions and interfaces appear cohesive.

Dashboard photo:

Launch & Delivery: March began with another round of client feedback (Client Review & Feedback #7). We swiftly implemented interface adjustments and bug fixes (UI Bug Fixes & Adjustments) (Final design photo). At this stage, we undertook the monumental task of reworking the AI Teacher into a fully offline version, which was successfully achieved. We also added significant new functionality – such as recording sessions and saving them during collaborative music creation in multiplayer mode.

Final design photo:

After another review (Client Review & Feedback #8), we incorporated the final recommendations, evaluated the overall volume of comments (Evaluate Feedback & Final Changes), and compiled the final backend (Compile Backend). In the final phase, we packaged the application (Package App) so that the client could easily deploy and maintain the product.

On March 4, team members Joyce and Ashton visited a local school to demonstrate Jamboxx Infinite to students and teachers. We received numerous comments about the exciting features of the application. Notably, an important request was made: to enable instrument playing by mouse hover rather than by click, making face navigation far more convenient. The children were especially impressed by the virtual instruments and minigames.

"It is incredibly rewarding to see that the product we are building truly finds a place in children's hearts. Their genuine smiles are the best motivation," said Joyce.

"I believe the feedback from the teachers was equally important. They know best what should be included in Jamboxx Infinite. There's still much work ahead, but after today, I feel more driven than ever," added Ashton.

Photo from the school:

On March 11, team members Kate and Wesley had the opportunity to present our project for 7 consecutive hours to various clients, partners, and school students. We started by demonstrating Jamboxx Infinite to high school students – many showed a keen interest. Most are studying Computer Science for their A-levels, so they were also curious about our project’s backend. The children enthusiastically tried the motion input for playing virtual instruments and competed in minigames.

"I was a bit nervous before the presentation, but all that quickly faded once I saw the genuine interest from the students. They asked questions not because they were prompted, but because they were truly curious and would love to use our project in the future," said Wesley.

"I was most inspired by hearing the children express their admiration for our mission. They said they aspire to become Computer Scientists and change the world for the better, just as we do. After such encouraging words, I felt completely at ease during our presentation to clients and industry partners," added Kate.

On March 25, we held the final presentation of the application before UCL professors. We received the final comments and recommendations, which was crucial given that only 2 days remained until final delivery to the client.

"I thought we were finished, but we received a lot of last-minute feedback that had to be urgently implemented. So there's no time to rest – we must keep working," said Joyce.