In latest Karakan v6.2 and folksKJ v4.30, they support a new format which allows playing non-ASCII lyrics on platforms with any language.
What's happening is when Karakan creates a song it always use a track with the track name "Soft karaoke" to store additional information, language, artist etc.
The keyword "@C" (in the information track) is used to specify the coding of the text. The code is only for one single track whose name is specified in the square brackets. By default, Karakan will generate following text event if Unicode option is enabled:
@C[Words]UTF-16LEThat means the text in the track "Words" is coded in UTF-16LE.
What to do When a player loads the song:
1) Find the track "Soft karaoke".
2) Get the information from "@C" keyword.
3) Follow the the track name in the square brackets.
4) Interprete the text in that track accordingly.
Karakan also allows editing song from right-to-left.
Currently, folksKJ is using the language key word "@L" to decide playing lyrics from right-to-left. Therefore it has to be added manually in Karakan. Following keywords are supported in folksKJ:
@LHEBR and
@LYIDDScreenshots from folksKJ:

http://sites.google.com/site/karakankaraoke/http://http://folksfun.com/Many Chinese songs in unicode format can be found in:
http://launch.groups.yahoo.com/group/softkaraoke/files/Songs/UNICODE%20KAR%20Files/