Make the voice generator correctly adjust itself for the current sound system configuration
Detail: s/Percussion - Similar to StringLib, Percussion uses various accumulator values that were tuned for a 48us sample period an 100Hz buffer fill rate. These values will now be calculated dynamically based around the current sound system configuration, ensuring the correct output is produced across all sensible configurations. Also swapped single-reg LDM/STM for Pull/Push to fix performance warnings from the assembler. Admin: Tested on BB-xM Voices now sound correct at high and low sample rates, small and large buffer sizes Part of the fix for ticket #347: https://www.riscosopen.org/tracker/tickets/347 Version 1.16. Tagged as 'Percussion-1_16'
Showing with 150 additions and 60 deletions