Steem Authors wrote:1) As far as I understand these things, we'd have to do fast Fourier transforms on the samples, which would complicate and slow down the sound processing;
Please put me straight if any of these points are wrong. As things stand, we're not planning to implement it.
Reading the specs of the LMC 1992 there is no need for any fourier trandforms to emulate the chip. The chip implements a highpass and lowpass filterr. Nothing that can't be solved by implementing some simple exponential filters:
new_output=(1-a)*old_output + a*input with a the 'filter value'
To give exact numbers and formula's I would have to look at the exact implementation of the LMC in the STE. I expect some very simple formula's will completely describe the filtering capabilities and volume and fading functions of the chip. The other part is the statemachine for the serial communication functions of the chip but that is not hard either (compared with the rest of the STe).