Ok, so some detailed how-its-made if anyone is interested.
Programs used:
Blender : for modeling,
Milkytracker : for the tune, with hand edited sample
yes. only a single one
Audacity : for the sound recording, editing
Gimp, mtpaint : for the textures, and the font
obj2tri.php : a piece of php code i am writing, that converts OBJ files to a JSON format that my "engine" can load
Backstory:
The low fidelity of the graphics is by design. As my target platform was my
Alcatel OT Fire phone (Firefox OS), which is an entry level piece of hw (1Ghz single core, Adreno 200 GPU). It lags a little bit, so there is some need for optimization, but it is watchable. This is also the cause of the 480px width of the canvas, so it fits on the 320x480 screen of the phone.
Here is a youtube video of it running on the phone :
http://youtu.be/Ez3E8pzLfEk (sorry for the whistle in the audio, and my webcam lags a bit more than the engine

)