Single word changed in one of the properties, derp.
This update was also to get this wallpaper included in the "Customizable" category. Why it wasn't in there from the start? ¯\_(ツ)_/¯

*Fixed a bug where the 'Direct' path solving algorithm would not work and be forever stuck doing nothing
*Some minor code optimizations

New:
*Added the consistency property. This property aims to make the mazes more realistic. The creation of the maze will prioritize a straight line N times more then the other available paths.

Fixed a bug where the maze only responded to audio when it felt like it

Bugs solved:
*There probably were some I noticed and fixed, but I forgot what they were... ¯\_(ツ)_/¯

Modified:
*The maze no longer updates along with each frame. Now there's two separate functions that handle the maze update (property speed) and FPS (regular wallpaper engine FPS limit)
This allows you to, for example, slow down the maze generation without compromising the audio-response smoothness
*A good part of the code was rewritten for better performance and to accomodate the new options

New:
*The program now respects the wallpaper engine FPS limit
*There's a new property: volume multiplier

New:
*Audio responsiveness is now available!! You can also chose how fast the maze is updated with the audio levels for better performance

Modified:
*The default wall color is now the same as the background

New:
*There's now a transition when the maze is about to get rebuilt

Bugs solved:
*The maze now should not initialize twice with different properties

Modified:
*Some changes to the code structure

New:
*New properties!! You can now:
-Select how long you want to wait before creating a new maze
-Chose not to solve the maze after it's built
-Change se seeking line behavior to random, best guess or direct

Modified:
*The grid no longer is recalculated from scratch everytime the maze restarts. Just when a property that requires recalculation changes. This should result in a lower average performance hit.

Bugs solved:
*The first and last element of the seeking line now respect the margins
*There should be much less (if any) pixel-level line artifacts in the walls

Modified:
*Infinitesimal performance improvement
*The way margins are handled changed. Instead of always creating a canvas the size of the screen and emulating margins within it, now the canvas size is the size left after removing all four margins and the canvas is moved around with CSS. This solved the first bug. I should've thought of this from the beginning...

New:
*Now the seeking line in bifurcations prioritizes the option that is closer to the goal in a straight line. Should make them look a little less stupid