Thanks i am glad that you like it, it's always nice when someone wants to use the programs.
i think one of the reason why the good low latency is.
the hole player routine is made in pure Asm,
and the main part that runs 2 million times i sec, and the part that make the sample buffer is highly optimize.
But the 3 timer and the score decoding could use a little optimization to run optimal,
but the CPU is not getting any slower so who cares
but i am thinking of making an extended version, that can run my old dos-score, so i will need a least 9 voices, or 18 voices.
with greater volume resolutions, different noise and hardware buzzer in each channels, so i will properly have to go back and optimize the hole deal.
/Peter