/* ======================================== * Crystal - Crystal.h * Copyright (c) 2016 airwindows, All rights reserved * ======================================== */ #ifndef __Crystal_H #include "Crystal.h" #endif void Crystal::processReplacing(float **inputs, float **outputs, VstInt32 sampleFrames) { float* in1 = inputs[0]; float* in2 = inputs[1]; float* out1 = outputs[0]; float* out2 = outputs[1]; double threshold = A; double hardness; double breakup = (1.0-(threshold/2.0))*3.14159265358979; double bridgerectifier; double sqdrive = B; if (sqdrive > 1.0) sqdrive *= sqdrive; sqdrive = sqrt(sqdrive); double indrive = C*3.0; if (indrive > 1.0) indrive *= indrive; indrive *= (1.0-(0.1695*sqdrive)); //no gain loss of convolution for APIcolypse //calibrate this to match noise level with character at 1.0 //you get for instance 0.819 and 1.0-0.819 is 0.181 double randy; double outlevel = D; if (threshold < 1) hardness = 1.0 / (1.0-threshold); else hardness = 999999999999999999999.0; //set up hardness to exactly fill gap between threshold and 0db //if threshold is literally 1 then hardness is infinite, so we make it very big while (--sampleFrames >= 0) { long double inputSampleL = *in1; long double inputSampleR = *in2; static int noisesourceL = 0; static int noisesourceR = 850010; int residue; double applyresidue; noisesourceL = noisesourceL % 1700021; noisesourceL++; residue = noisesourceL * noisesourceL; residue = residue % 170003; residue *= residue; residue = residue % 17011; residue *= residue; residue = residue % 1709; residue *= residue; residue = residue % 173; residue *= residue; residue = residue % 17; applyresidue = residue; applyresidue *= 0.00000001; applyresidue *= 0.00000001; inputSampleL += applyresidue; if (inputSampleL<1.2e-38 && -inputSampleL<1.2e-38) { inputSampleL -= applyresidue; } noisesourceR = noisesourceR % 1700021; noisesourceR++; residue = noisesourceR * noisesourceR; residue = residue % 170003; residue *= residue; residue = residue % 17011; residue *= residue; residue = residue % 1709; residue *= residue; residue = residue % 173; residue *= residue; residue = residue % 17; applyresidue = residue; applyresidue *= 0.00000001; applyresidue *= 0.00000001; inputSampleR += applyresidue; if (inputSampleR<1.2e-38 && -inputSampleR<1.2e-38) { inputSampleR -= applyresidue; } //for live air, we always apply the dither noise. Then, if our result is //effectively digital black, we'll subtract it again. We want a 'air' hiss inputSampleL *= indrive; inputSampleR *= indrive; //calibrated to match gain through convolution and -0.3 correction if (sqdrive > 0.0){ bL[23] = bL[22]; bL[22] = bL[21]; bL[21] = bL[20]; bL[20] = bL[19]; bL[19] = bL[18]; bL[18] = bL[17]; bL[17] = bL[16]; bL[16] = bL[15]; bL[15] = bL[14]; bL[14] = bL[13]; bL[13] = bL[12]; bL[12] = bL[11]; bL[11] = bL[10]; bL[10] = bL[9]; bL[9] = bL[8]; bL[8] = bL[7]; bL[7] = bL[6]; bL[6] = bL[5]; bL[5] = bL[4]; bL[4] = bL[3]; bL[3] = bL[2]; bL[2] = bL[1]; bL[1] = bL[0]; bL[0] = inputSampleL * sqdrive; inputSampleL += (bL[1] * (0.38856694371895023 + (0.14001177830115491*fabs(bL[1])))); inputSampleL -= (bL[2] * (0.17469488984546111 + (0.05204541941091459*fabs(bL[2])))); inputSampleL += (bL[3] * (0.11643521461774288 - (0.01193121216518472*fabs(bL[3])))); inputSampleL -= (bL[4] * (0.08874416268268183 - (0.05867502375036486*fabs(bL[4])))); inputSampleL += (bL[5] * (0.07222999223073785 - (0.08519974113692971*fabs(bL[5])))); inputSampleL -= (bL[6] * (0.06103207678880003 - (0.09230674983449150*fabs(bL[6])))); inputSampleL += (bL[7] * (0.05277389277465404 - (0.08487342372497046*fabs(bL[7])))); inputSampleL -= (bL[8] * (0.04631144388636078 - (0.06976851898821038*fabs(bL[8])))); inputSampleL += (bL[9] * (0.04102721072495113 - (0.05337974329110802*fabs(bL[9])))); inputSampleL -= (bL[10] * (0.03656047655964371 - (0.03990914278458497*fabs(bL[10])))); inputSampleL += (bL[11] * (0.03268677450573373 - (0.03090433934018759*fabs(bL[11])))); inputSampleL -= (bL[12] * (0.02926012259262895 - (0.02585223214266682*fabs(bL[12])))); inputSampleL += (bL[13] * (0.02618257163789973 - (0.02326667039588473*fabs(bL[13])))); inputSampleL -= (bL[14] * (0.02338568277879992 - (0.02167067760829789*fabs(bL[14])))); inputSampleL += (bL[15] * (0.02082142324645262 - (0.02013392273267951*fabs(bL[15])))); inputSampleL -= (bL[16] * (0.01845525966656259 - (0.01833038930966512*fabs(bL[16])))); inputSampleL += (bL[17] * (0.01626113504980445 - (0.01631893218593511*fabs(bL[17])))); inputSampleL -= (bL[18] * (0.01422084088669267 - (0.01427828125219885*fabs(bL[18])))); inputSampleL += (bL[19] * (0.01231993595709338 - (0.01233991521342998*fabs(bL[19])))); inputSampleL -= (bL[20] * (0.01054774630451013 - (0.01054774630542346*fabs(bL[20])))); inputSampleL += (bL[21] * (0.00889548162355088 - (0.00889548162263755*fabs(bL[21])))); inputSampleL -= (bL[22] * (0.00735749099304526 - (0.00735749099395860*fabs(bL[22])))); inputSampleL += (bL[23] * (0.00592812350468000 - (0.00592812350376666*fabs(bL[23])))); } //the Character plugins as individual processors did this. BussColors applies an averaging factor to produce // more of a consistent variation between soft and loud convolutions. For years I thought this code was a //mistake and did nothing, but in fact what it's doing is producing slightly different curves for every single //convolution kernel location: this will be true of the Character individual plugins as well. //calibrated to match gain through convolution and -0.3 correction if (sqdrive > 0.0){ bR[23] = bR[22]; bR[22] = bR[21]; bR[21] = bR[20]; bR[20] = bR[19]; bR[19] = bR[18]; bR[18] = bR[17]; bR[17] = bR[16]; bR[16] = bR[15]; bR[15] = bR[14]; bR[14] = bR[13]; bR[13] = bR[12]; bR[12] = bR[11]; bR[11] = bR[10]; bR[10] = bR[9]; bR[9] = bR[8]; bR[8] = bR[7]; bR[7] = bR[6]; bR[6] = bR[5]; bR[5] = bR[4]; bR[4] = bR[3]; bR[3] = bR[2]; bR[2] = bR[1]; bR[1] = bR[0]; bR[0] = inputSampleR * sqdrive; inputSampleR += (bR[1] * (0.38856694371895023 + (0.14001177830115491*fabs(bR[1])))); inputSampleR -= (bR[2] * (0.17469488984546111 + (0.05204541941091459*fabs(bR[2])))); inputSampleR += (bR[3] * (0.11643521461774288 - (0.01193121216518472*fabs(bR[3])))); inputSampleR -= (bR[4] * (0.08874416268268183 - (0.05867502375036486*fabs(bR[4])))); inputSampleR += (bR[5] * (0.07222999223073785 - (0.08519974113692971*fabs(bR[5])))); inputSampleR -= (bR[6] * (0.06103207678880003 - (0.09230674983449150*fabs(bR[6])))); inputSampleR += (bR[7] * (0.05277389277465404 - (0.08487342372497046*fabs(bR[7])))); inputSampleR -= (bR[8] * (0.04631144388636078 - (0.06976851898821038*fabs(bR[8])))); inputSampleR += (bR[9] * (0.04102721072495113 - (0.05337974329110802*fabs(bR[9])))); inputSampleR -= (bR[10] * (0.03656047655964371 - (0.03990914278458497*fabs(bR[10])))); inputSampleR += (bR[11] * (0.03268677450573373 - (0.03090433934018759*fabs(bR[11])))); inputSampleR -= (bR[12] * (0.02926012259262895 - (0.02585223214266682*fabs(bR[12])))); inputSampleR += (bR[13] * (0.02618257163789973 - (0.02326667039588473*fabs(bR[13])))); inputSampleR -= (bR[14] * (0.02338568277879992 - (0.02167067760829789*fabs(bR[14])))); inputSampleR += (bR[15] * (0.02082142324645262 - (0.02013392273267951*fabs(bR[15])))); inputSampleR -= (bR[16] * (0.01845525966656259 - (0.01833038930966512*fabs(bR[16])))); inputSampleR += (bR[17] * (0.01626113504980445 - (0.01631893218593511*fabs(bR[17])))); inputSampleR -= (bR[18] * (0.01422084088669267 - (0.01427828125219885*fabs(bR[18])))); inputSampleR += (bR[19] * (0.01231993595709338 - (0.01233991521342998*fabs(bR[19])))); inputSampleR -= (bR[20] * (0.01054774630451013 - (0.01054774630542346*fabs(bR[20])))); inputSampleR += (bR[21] * (0.00889548162355088 - (0.00889548162263755*fabs(bR[21])))); inputSampleR -= (bR[22] * (0.00735749099304526 - (0.00735749099395860*fabs(bR[22])))); inputSampleR += (bR[23] * (0.00592812350468000 - (0.00592812350376666*fabs(bR[23])))); } //the Character plugins as individual processors did this. BussColors applies an averaging factor to produce // more of a consistent variation between soft and loud convolutions. For years I thought this code was a //mistake and did nothing, but in fact what it's doing is producing slightly different curves for every single //convolution kernel location: this will be true of the Character individual plugins as well. if (fabs(inputSampleL) > threshold) { bridgerectifier = (fabs(inputSampleL)-threshold)*hardness; //skip flat area if any, scale to distortion limit if (bridgerectifier > breakup) bridgerectifier = breakup; //max value for sine function, 'breakup' modeling for trashed console tone //more hardness = more solidness behind breakup modeling. more softness, more 'grunge' and sag bridgerectifier = sin(bridgerectifier)/hardness; //do the sine factor, scale back to proper amount if (inputSampleL > 0) inputSampleL = bridgerectifier+threshold; else inputSampleL = -(bridgerectifier+threshold); } //otherwise we leave it untouched by the overdrive stuff //this is the notorious New Channel Density algorithm. It's much less popular than the original Density, //because it introduces a point where the saturation 'curve' changes from straight to curved. //People don't like these discontinuities, but you can use them for effect or to grit up the sound. if (fabs(inputSampleR) > threshold) { bridgerectifier = (fabs(inputSampleR)-threshold)*hardness; //skip flat area if any, scale to distortion limit if (bridgerectifier > breakup) bridgerectifier = breakup; //max value for sine function, 'breakup' modeling for trashed console tone //more hardness = more solidness behind breakup modeling. more softness, more 'grunge' and sag bridgerectifier = sin(bridgerectifier)/hardness; //do the sine factor, scale back to proper amount if (inputSampleR > 0) inputSampleR = bridgerectifier+threshold; else inputSampleR = -(bridgerectifier+threshold); } //otherwise we leave it untouched by the overdrive stuff //this is the notorious New Channel Density algorithm. It's much less popular than the original Density, //because it introduces a point where the saturation 'curve' changes from straight to curved. //People don't like these discontinuities, but you can use them for effect or to grit up the sound. randy = ((rand()/(double)RAND_MAX)*0.022); bridgerectifier = ((inputSampleL*(1-randy))+(lastSampleL*randy)) * outlevel; lastSampleL = inputSampleL; inputSampleL = bridgerectifier; //applies a tiny 'fuzz' to highs: from original Crystal. randy = ((rand()/(double)RAND_MAX)*0.022); bridgerectifier = ((inputSampleR*(1-randy))+(lastSampleR*randy)) * outlevel; lastSampleR = inputSampleR; inputSampleR = bridgerectifier; //applies a tiny 'fuzz' to highs: from original Crystal. //This is akin to the old Chrome Oxide plugin, applying a fuzz to only the slews. The noise only appears //when current and old samples are different from each other, otherwise you can't tell it's there. //This is not only during silence but the tops of low frequency waves: it scales down to affect lows more gently. //stereo 32 bit dither, made small and tidy. int expon; frexpf((float)inputSampleL, &expon); long double dither = (rand()/(RAND_MAX*7.737125245533627e+25))*pow(2,expon+62); inputSampleL += (dither-fpNShapeL); fpNShapeL = dither; frexpf((float)inputSampleR, &expon); dither = (rand()/(RAND_MAX*7.737125245533627e+25))*pow(2,expon+62); inputSampleR += (dither-fpNShapeR); fpNShapeR = dither; //end 32 bit dither *out1 = inputSampleL; *out2 = inputSampleR; *in1++; *in2++; *out1++; *out2++; } } void Crystal::processDoubleReplacing(double **inputs, double **outputs, VstInt32 sampleFrames) { double* in1 = inputs[0]; double* in2 = inputs[1]; double* out1 = outputs[0]; double* out2 = outputs[1]; double threshold = A; double hardness; double breakup = (1.0-(threshold/2.0))*3.14159265358979; double bridgerectifier; double sqdrive = B; if (sqdrive > 1.0) sqdrive *= sqdrive; sqdrive = sqrt(sqdrive); double indrive = C*3.0; if (indrive > 1.0) indrive *= indrive; indrive *= (1.0-(0.1695*sqdrive)); //no gain loss of convolution for APIcolypse //calibrate this to match noise level with character at 1.0 //you get for instance 0.819 and 1.0-0.819 is 0.181 double randy; double outlevel = D; if (threshold < 1) hardness = 1.0 / (1.0-threshold); else hardness = 999999999999999999999.0; //set up hardness to exactly fill gap between threshold and 0db //if threshold is literally 1 then hardness is infinite, so we make it very big while (--sampleFrames >= 0) { long double inputSampleL = *in1; long double inputSampleR = *in2; static int noisesourceL = 0; static int noisesourceR = 850010; int residue; double applyresidue; noisesourceL = noisesourceL % 1700021; noisesourceL++; residue = noisesourceL * noisesourceL; residue = residue % 170003; residue *= residue; residue = residue % 17011; residue *= residue; residue = residue % 1709; residue *= residue; residue = residue % 173; residue *= residue; residue = residue % 17; applyresidue = residue; applyresidue *= 0.00000001; applyresidue *= 0.00000001; inputSampleL += applyresidue; if (inputSampleL<1.2e-38 && -inputSampleL<1.2e-38) { inputSampleL -= applyresidue; } noisesourceR = noisesourceR % 1700021; noisesourceR++; residue = noisesourceR * noisesourceR; residue = residue % 170003; residue *= residue; residue = residue % 17011; residue *= residue; residue = residue % 1709; residue *= residue; residue = residue % 173; residue *= residue; residue = residue % 17; applyresidue = residue; applyresidue *= 0.00000001; applyresidue *= 0.00000001; inputSampleR += applyresidue; if (inputSampleR<1.2e-38 && -inputSampleR<1.2e-38) { inputSampleR -= applyresidue; } //for live air, we always apply the dither noise. Then, if our result is //effectively digital black, we'll subtract it again. We want a 'air' hiss inputSampleL *= indrive; inputSampleR *= indrive; //calibrated to match gain through convolution and -0.3 correction if (sqdrive > 0.0){ bL[23] = bL[22]; bL[22] = bL[21]; bL[21] = bL[20]; bL[20] = bL[19]; bL[19] = bL[18]; bL[18] = bL[17]; bL[17] = bL[16]; bL[16] = bL[15]; bL[15] = bL[14]; bL[14] = bL[13]; bL[13] = bL[12]; bL[12] = bL[11]; bL[11] = bL[10]; bL[10] = bL[9]; bL[9] = bL[8]; bL[8] = bL[7]; bL[7] = bL[6]; bL[6] = bL[5]; bL[5] = bL[4]; bL[4] = bL[3]; bL[3] = bL[2]; bL[2] = bL[1]; bL[1] = bL[0]; bL[0] = inputSampleL * sqdrive; inputSampleL += (bL[1] * (0.38856694371895023 + (0.14001177830115491*fabs(bL[1])))); inputSampleL -= (bL[2] * (0.17469488984546111 + (0.05204541941091459*fabs(bL[2])))); inputSampleL += (bL[3] * (0.11643521461774288 - (0.01193121216518472*fabs(bL[3])))); inputSampleL -= (bL[4] * (0.08874416268268183 - (0.05867502375036486*fabs(bL[4])))); inputSampleL += (bL[5] * (0.07222999223073785 - (0.08519974113692971*fabs(bL[5])))); inputSampleL -= (bL[6] * (0.06103207678880003 - (0.09230674983449150*fabs(bL[6])))); inputSampleL += (bL[7] * (0.05277389277465404 - (0.08487342372497046*fabs(bL[7])))); inputSampleL -= (bL[8] * (0.04631144388636078 - (0.06976851898821038*fabs(bL[8])))); inputSampleL += (bL[9] * (0.04102721072495113 - (0.05337974329110802*fabs(bL[9])))); inputSampleL -= (bL[10] * (0.03656047655964371 - (0.03990914278458497*fabs(bL[10])))); inputSampleL += (bL[11] * (0.03268677450573373 - (0.03090433934018759*fabs(bL[11])))); inputSampleL -= (bL[12] * (0.02926012259262895 - (0.02585223214266682*fabs(bL[12])))); inputSampleL += (bL[13] * (0.02618257163789973 - (0.02326667039588473*fabs(bL[13])))); inputSampleL -= (bL[14] * (0.02338568277879992 - (0.02167067760829789*fabs(bL[14])))); inputSampleL += (bL[15] * (0.02082142324645262 - (0.02013392273267951*fabs(bL[15])))); inputSampleL -= (bL[16] * (0.01845525966656259 - (0.01833038930966512*fabs(bL[16])))); inputSampleL += (bL[17] * (0.01626113504980445 - (0.01631893218593511*fabs(bL[17])))); inputSampleL -= (bL[18] * (0.01422084088669267 - (0.01427828125219885*fabs(bL[18])))); inputSampleL += (bL[19] * (0.01231993595709338 - (0.01233991521342998*fabs(bL[19])))); inputSampleL -= (bL[20] * (0.01054774630451013 - (0.01054774630542346*fabs(bL[20])))); inputSampleL += (bL[21] * (0.00889548162355088 - (0.00889548162263755*fabs(bL[21])))); inputSampleL -= (bL[22] * (0.00735749099304526 - (0.00735749099395860*fabs(bL[22])))); inputSampleL += (bL[23] * (0.00592812350468000 - (0.00592812350376666*fabs(bL[23])))); } //the Character plugins as individual processors did this. BussColors applies an averaging factor to produce // more of a consistent variation between soft and loud convolutions. For years I thought this code was a //mistake and did nothing, but in fact what it's doing is producing slightly different curves for every single //convolution kernel location: this will be true of the Character individual plugins as well. //calibrated to match gain through convolution and -0.3 correction if (sqdrive > 0.0){ bR[23] = bR[22]; bR[22] = bR[21]; bR[21] = bR[20]; bR[20] = bR[19]; bR[19] = bR[18]; bR[18] = bR[17]; bR[17] = bR[16]; bR[16] = bR[15]; bR[15] = bR[14]; bR[14] = bR[13]; bR[13] = bR[12]; bR[12] = bR[11]; bR[11] = bR[10]; bR[10] = bR[9]; bR[9] = bR[8]; bR[8] = bR[7]; bR[7] = bR[6]; bR[6] = bR[5]; bR[5] = bR[4]; bR[4] = bR[3]; bR[3] = bR[2]; bR[2] = bR[1]; bR[1] = bR[0]; bR[0] = inputSampleR * sqdrive; inputSampleR += (bR[1] * (0.38856694371895023 + (0.14001177830115491*fabs(bR[1])))); inputSampleR -= (bR[2] * (0.17469488984546111 + (0.05204541941091459*fabs(bR[2])))); inputSampleR += (bR[3] * (0.11643521461774288 - (0.01193121216518472*fabs(bR[3])))); inputSampleR -= (bR[4] * (0.08874416268268183 - (0.05867502375036486*fabs(bR[4])))); inputSampleR += (bR[5] * (0.07222999223073785 - (0.08519974113692971*fabs(bR[5])))); inputSampleR -= (bR[6] * (0.06103207678880003 - (0.09230674983449150*fabs(bR[6])))); inputSampleR += (bR[7] * (0.05277389277465404 - (0.08487342372497046*fabs(bR[7])))); inputSampleR -= (bR[8] * (0.04631144388636078 - (0.06976851898821038*fabs(bR[8])))); inputSampleR += (bR[9] * (0.04102721072495113 - (0.05337974329110802*fabs(bR[9])))); inputSampleR -= (bR[10] * (0.03656047655964371 - (0.03990914278458497*fabs(bR[10])))); inputSampleR += (bR[11] * (0.03268677450573373 - (0.03090433934018759*fabs(bR[11])))); inputSampleR -= (bR[12] * (0.02926012259262895 - (0.02585223214266682*fabs(bR[12])))); inputSampleR += (bR[13] * (0.02618257163789973 - (0.02326667039588473*fabs(bR[13])))); inputSampleR -= (bR[14] * (0.02338568277879992 - (0.02167067760829789*fabs(bR[14])))); inputSampleR += (bR[15] * (0.02082142324645262 - (0.02013392273267951*fabs(bR[15])))); inputSampleR -= (bR[16] * (0.01845525966656259 - (0.01833038930966512*fabs(bR[16])))); inputSampleR += (bR[17] * (0.01626113504980445 - (0.01631893218593511*fabs(bR[17])))); inputSampleR -= (bR[18] * (0.01422084088669267 - (0.01427828125219885*fabs(bR[18])))); inputSampleR += (bR[19] * (0.01231993595709338 - (0.01233991521342998*fabs(bR[19])))); inputSampleR -= (bR[20] * (0.01054774630451013 - (0.01054774630542346*fabs(bR[20])))); inputSampleR += (bR[21] * (0.00889548162355088 - (0.00889548162263755*fabs(bR[21])))); inputSampleR -= (bR[22] * (0.00735749099304526 - (0.00735749099395860*fabs(bR[22])))); inputSampleR += (bR[23] * (0.00592812350468000 - (0.00592812350376666*fabs(bR[23])))); } //the Character plugins as individual processors did this. BussColors applies an averaging factor to produce // more of a consistent variation between soft and loud convolutions. For years I thought this code was a //mistake and did nothing, but in fact what it's doing is producing slightly different curves for every single //convolution kernel location: this will be true of the Character individual plugins as well. if (fabs(inputSampleL) > threshold) { bridgerectifier = (fabs(inputSampleL)-threshold)*hardness; //skip flat area if any, scale to distortion limit if (bridgerectifier > breakup) bridgerectifier = breakup; //max value for sine function, 'breakup' modeling for trashed console tone //more hardness = more solidness behind breakup modeling. more softness, more 'grunge' and sag bridgerectifier = sin(bridgerectifier)/hardness; //do the sine factor, scale back to proper amount if (inputSampleL > 0) inputSampleL = bridgerectifier+threshold; else inputSampleL = -(bridgerectifier+threshold); } //otherwise we leave it untouched by the overdrive stuff //this is the notorious New Channel Density algorithm. It's much less popular than the original Density, //because it introduces a point where the saturation 'curve' changes from straight to curved. //People don't like these discontinuities, but you can use them for effect or to grit up the sound. if (fabs(inputSampleR) > threshold) { bridgerectifier = (fabs(inputSampleR)-threshold)*hardness; //skip flat area if any, scale to distortion limit if (bridgerectifier > breakup) bridgerectifier = breakup; //max value for sine function, 'breakup' modeling for trashed console tone //more hardness = more solidness behind breakup modeling. more softness, more 'grunge' and sag bridgerectifier = sin(bridgerectifier)/hardness; //do the sine factor, scale back to proper amount if (inputSampleR > 0) inputSampleR = bridgerectifier+threshold; else inputSampleR = -(bridgerectifier+threshold); } //otherwise we leave it untouched by the overdrive stuff //this is the notorious New Channel Density algorithm. It's much less popular than the original Density, //because it introduces a point where the saturation 'curve' changes from straight to curved. //People don't like these discontinuities, but you can use them for effect or to grit up the sound. randy = ((rand()/(double)RAND_MAX)*0.022); bridgerectifier = ((inputSampleL*(1-randy))+(lastSampleL*randy)) * outlevel; lastSampleL = inputSampleL; inputSampleL = bridgerectifier; //applies a tiny 'fuzz' to highs: from original Crystal. randy = ((rand()/(double)RAND_MAX)*0.022); bridgerectifier = ((inputSampleR*(1-randy))+(lastSampleR*randy)) * outlevel; lastSampleR = inputSampleR; inputSampleR = bridgerectifier; //applies a tiny 'fuzz' to highs: from original Crystal. //This is akin to the old Chrome Oxide plugin, applying a fuzz to only the slews. The noise only appears //when current and old samples are different from each other, otherwise you can't tell it's there. //This is not only during silence but the tops of low frequency waves: it scales down to affect lows more gently. //stereo 64 bit dither, made small and tidy. int expon; frexp((double)inputSampleL, &expon); long double dither = (rand()/(RAND_MAX*7.737125245533627e+25))*pow(2,expon+62); dither /= 536870912.0; //needs this to scale to 64 bit zone inputSampleL += (dither-fpNShapeL); fpNShapeL = dither; frexp((double)inputSampleR, &expon); dither = (rand()/(RAND_MAX*7.737125245533627e+25))*pow(2,expon+62); dither /= 536870912.0; //needs this to scale to 64 bit zone inputSampleR += (dither-fpNShapeR); fpNShapeR = dither; //end 64 bit dither *out1 = inputSampleL; *out2 = inputSampleR; *in1++; *in2++; *out1++; *out2++; } }