/* ========================================
 *  ToTape9 - ToTape9.h
 *  Copyright (c) airwindows, Airwindows uses the MIT license
 * ======================================== */

#ifndef __ToTape9_H
#include "ToTape9.h"
#endif

void ToTape9::processReplacing(float **inputs, float **outputs, VstInt32 sampleFrames) 
{
    float* in1  =  inputs[0];
    float* in2  =  inputs[1];
    float* out1 = outputs[0];
    float* out2 = outputs[1];

	double overallscale = 1.0;
	overallscale /= 44100.0;
	overallscale *= getSampleRate();
	
	int spacing = floor(overallscale); //should give us working basic scaling, usually 2 or 4
	if (spacing < 1) spacing = 1; if (spacing > 16) spacing = 16;
	int slewsing = floor(overallscale*2.0);
	if (slewsing < 2) slewsing = 2; if (slewsing > 32) slewsing = 32;
	
	double inputGain = pow(A*2.0,2.0);
	
	double dublyAmount = B*2.0;
	double outlyAmount = (1.0-B)*-2.0;
	if (outlyAmount < -1.0) outlyAmount = -1.0;
	double iirEncFreq = (1.0-C)/overallscale;
	double iirDecFreq = C/overallscale;
	
	double flutDepth = pow(D,6)*overallscale*50;
	if (flutDepth > 498.0) flutDepth = 498.0;
	double flutFrequency = (0.02*pow(E,3))/overallscale;
	double bias = (F*2.0)-1.0;
	double underBias = (pow(bias,4)*0.25)/overallscale;
	double overBias = pow(1.0-bias,3)/overallscale;
	if (bias > 0.0) underBias = 0.0;
	if (bias < 0.0) overBias = 1.0/overallscale;
	
	gslew[threshold9] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold8] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold7] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold6] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold5] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold4] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold3] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold2] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold1] = overBias;
	overBias *= 1.618033988749894848204586;
	
	double headBumpDrive = (G*0.1)/overallscale;
	double headBumpMix = G*0.5;
	
	hdbA[hdb_freq] = (((H*H)*175.0)+25.0)/getSampleRate();
	hdbB[hdb_freq] = hdbA[hdb_freq]*0.9375;
	hdbB[hdb_reso] = hdbA[hdb_reso] = 0.618033988749894848204586;
	hdbB[hdb_a1] = hdbA[hdb_a1] = 0.0;
	
	double K = tan(M_PI * hdbA[hdb_freq]);
	double norm = 1.0 / (1.0 + K / hdbA[hdb_reso] + K * K);
	hdbA[hdb_a0] = K / hdbA[hdb_reso] * norm;
	hdbA[hdb_a2] = -hdbA[hdb_a0];
	hdbA[hdb_b1] = 2.0 * (K * K - 1.0) * norm;
	hdbA[hdb_b2] = (1.0 - K / hdbA[hdb_reso] + K * K) * norm;
	K = tan(M_PI * hdbB[hdb_freq]);
	norm = 1.0 / (1.0 + K / hdbB[hdb_reso] + K * K);
	hdbB[hdb_a0] = K / hdbB[hdb_reso] * norm;
	hdbB[hdb_a2] = -hdbB[hdb_a0];
	hdbB[hdb_b1] = 2.0 * (K * K - 1.0) * norm;
	hdbB[hdb_b2] = (1.0 - K / hdbB[hdb_reso] + K * K) * norm;
	
	double outputGain = I*2.0;
	
    while (--sampleFrames >= 0)
    {
		double inputSampleL = *in1;
		double inputSampleR = *in2;
		if (fabs(inputSampleL)<1.18e-23) inputSampleL = fpdL * 1.18e-17;
		if (fabs(inputSampleR)<1.18e-23) inputSampleR = fpdR * 1.18e-17;
		
		if (inputGain != 1.0) {
			inputSampleL *= inputGain;
			inputSampleR *= inputGain;
		}
		
		//Dubly encode
		iirEncL = (iirEncL * (1.0 - iirEncFreq)) + (inputSampleL * iirEncFreq);
		double highPart = ((inputSampleL-iirEncL)*2.848);
		highPart += avgEncL; avgEncL = (inputSampleL-iirEncL)*1.152;
		if (highPart > 1.0) highPart = 1.0; if (highPart < -1.0) highPart = -1.0;
		double dubly = fabs(highPart);
		if (dubly > 0.0) {
			double adjust = log(1.0+(255.0*dubly))/2.40823996531;
			if (adjust > 0.0) dubly /= adjust;
			compEncL = (compEncL*(1.0-iirEncFreq))+(dubly*iirEncFreq);
			inputSampleL += ((highPart*compEncL)*dublyAmount);
		} //end Dubly encode L
		iirEncR = (iirEncR * (1.0 - iirEncFreq)) + (inputSampleR * iirEncFreq);
		highPart = ((inputSampleR-iirEncR)*2.848);
		highPart += avgEncR; avgEncR = (inputSampleR-iirEncR)*1.152;
		if (highPart > 1.0) highPart = 1.0; if (highPart < -1.0) highPart = -1.0;
		dubly = fabs(highPart);
		if (dubly > 0.0) {
			double adjust = log(1.0+(255.0*dubly))/2.40823996531;
			if (adjust > 0.0) dubly /= adjust;
			compEncR = (compEncR*(1.0-iirEncFreq))+(dubly*iirEncFreq);
			inputSampleR += ((highPart*compEncR)*dublyAmount);
		} //end Dubly encode R
		
		//begin Flutter
		if (flutDepth > 0.0) {
			if (gcount < 0 || gcount > 999) gcount = 999;
			dL[gcount] = inputSampleL;
			int count = gcount;
			double offset = flutDepth + (flutDepth * sin(sweepL));
			sweepL += nextmaxL * flutFrequency;
			if (sweepL > (M_PI*2.0)) {
				sweepL -= M_PI*2.0;
				double flutA = 0.24 + (fpdL / (double)UINT32_MAX * 0.74);
				fpdL ^= fpdL << 13; fpdL ^= fpdL >> 17; fpdL ^= fpdL << 5;
				double flutB = 0.24 + (fpdL / (double)UINT32_MAX * 0.74);
				if (fabs(flutA-sin(sweepR+nextmaxR))<fabs(flutB-sin(sweepR+nextmaxR))) nextmaxL = flutA; else nextmaxL = flutB;
			}
			count += (int)floor(offset);
			inputSampleL = (dL[count-((count > 999)?1000:0)] * (1-(offset-floor(offset))));
			inputSampleL += (dL[count+1-((count+1 > 999)?1000:0)] * (offset-floor(offset)));
			dR[gcount] = inputSampleR;
			count = gcount;
			offset = flutDepth + (flutDepth * sin(sweepR));
			sweepR += nextmaxR * flutFrequency;
			if (sweepR > (M_PI*2.0)) {
				sweepR -= M_PI*2.0;
				double flutA = 0.24 + (fpdR / (double)UINT32_MAX * 0.74);
				fpdR ^= fpdR << 13; fpdR ^= fpdR >> 17; fpdR ^= fpdR << 5;
				double flutB = 0.24 + (fpdR / (double)UINT32_MAX * 0.74);
				if (fabs(flutA-sin(sweepL+nextmaxL))<fabs(flutB-sin(sweepL+nextmaxL))) nextmaxR = flutA; else nextmaxR = flutB;
			}
			count += (int)floor(offset);
			inputSampleR = (dR[count-((count > 999)?1000:0)] * (1-(offset-floor(offset))));
			inputSampleR += (dR[count+1-((count+1 > 999)?1000:0)] * (offset-floor(offset)));
			gcount--;
		}
		//end Flutter
		
		//start bias routine
		if (fabs(bias) > 0.001) {
			for (int x = 0; x < gslew_total; x += 3) {
				if (underBias > 0.0) {
					double stuck = fabs(inputSampleL - (gslew[x]/0.975)) / underBias;
					if (stuck < 1.0) inputSampleL = (inputSampleL * stuck) + ((gslew[x]/0.975)*(1.0-stuck));
					stuck =  fabs(inputSampleR - (gslew[x+1]/0.975)) / underBias;
					if (stuck < 1.0) inputSampleR = (inputSampleR * stuck) + ((gslew[x+1]/0.975)*(1.0-stuck));
				}
				if ((inputSampleL - gslew[x]) > gslew[x+2]) inputSampleL = gslew[x] + gslew[x+2];
				if (-(inputSampleL - gslew[x]) > gslew[x+2]) inputSampleL = gslew[x] - gslew[x+2];
				gslew[x] = inputSampleL * 0.975;
				if ((inputSampleR - gslew[x+1]) > gslew[x+2]) inputSampleR = gslew[x+1] + gslew[x+2];
				if (-(inputSampleR - gslew[x+1]) > gslew[x+2]) inputSampleR = gslew[x+1] - gslew[x+2];
				gslew[x+1] = inputSampleR * 0.975;
			}
		}
		//end bias routine
		
		//begin tiny hysteresis
		double applyHysteresis = (1.0-fabs(inputSampleL))*(1.0-fabs(inputSampleL))*0.012;
		hysteresisL = fmax(fmin(hysteresisL+((inputSampleL*fabs(inputSampleL))),0.011449),-0.011449)*0.999;
		inputSampleL += (hysteresisL*applyHysteresis);
		applyHysteresis = (1.0-fabs(inputSampleR))*(1.0-fabs(inputSampleR))*0.012;
		hysteresisR = fmax(fmin(hysteresisR+((inputSampleR*fabs(inputSampleR))),0.011449),-0.011449)*0.999;
		inputSampleR += (hysteresisR*applyHysteresis);
		
		//begin TapeHack2
		double darkSampleL = inputSampleL;
		double darkSampleR = inputSampleR;
		if (avgPos > 31) avgPos = 0;
		if (slewsing > 31) {
			avg32L[avgPos] = darkSampleL; avg32R[avgPos] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 32; x++) {darkSampleL += avg32L[x]; darkSampleR += avg32R[x];}
			darkSampleL /= 32.0; darkSampleR /= 32.0;
		} if (slewsing > 15) {
			avg16L[avgPos%16] = darkSampleL; avg16R[avgPos%16] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 16; x++) {darkSampleL += avg16L[x]; darkSampleR += avg16R[x];}
			darkSampleL /= 16.0; darkSampleR /= 16.0;
		} if (slewsing > 7) {
			avg8L[avgPos%8] = darkSampleL; avg8R[avgPos%8] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 8; x++) {darkSampleL += avg8L[x]; darkSampleR += avg8R[x];}
			darkSampleL /= 8.0; darkSampleR /= 8.0;
		} if (slewsing > 3) {
			avg4L[avgPos%4] = darkSampleL; avg4R[avgPos%4] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 4; x++) {darkSampleL += avg4L[x]; darkSampleR += avg4R[x];}
			darkSampleL /= 4.0; darkSampleR /= 4.0;
		} if (slewsing > 1) {
			avg2L[avgPos%2] = darkSampleL; avg2R[avgPos%2] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 2; x++) {darkSampleL += avg2L[x]; darkSampleR += avg2R[x];}
			darkSampleL /= 2.0; darkSampleR /= 2.0; 
		} //only update avgPos after the post-distortion filter stage
		double avgSlewL = fmin(fabs(lastDarkL-inputSampleL)*0.12*overallscale,1.0);
		avgSlewL = 1.0-(1.0-avgSlewL*1.0-avgSlewL);
		inputSampleL = (inputSampleL*(1.0-avgSlewL)) + (darkSampleL*avgSlewL);
		lastDarkL = darkSampleL;
		double avgSlewR = fmin(fabs(lastDarkR-inputSampleR)*0.12*overallscale,1.0);
		avgSlewR = 1.0-(1.0-avgSlewR*1.0-avgSlewR);
		inputSampleR = (inputSampleR*(1.0-avgSlewR)) + (darkSampleR*avgSlewR);
		lastDarkR = darkSampleR;
		
		//begin TapeHack
		inputSampleL = fmax(fmin(inputSampleL,2.305929007734908),-2.305929007734908);
		double addtwo = inputSampleL * inputSampleL;
		double empower = inputSampleL * addtwo; // inputSample to the third power
		inputSampleL -= (empower / 6.0);
		empower *= addtwo; // to the fifth power
		inputSampleL += (empower / 69.0);
		empower *= addtwo; //seventh
		inputSampleL -= (empower / 2530.08);
		empower *= addtwo; //ninth
		inputSampleL += (empower / 224985.6);
		empower *= addtwo; //eleventh
		inputSampleL -= (empower / 9979200.0f);
		//this is a degenerate form of a Taylor Series to approximate sin()
		
		inputSampleR = fmax(fmin(inputSampleR,2.305929007734908),-2.305929007734908);
		addtwo = inputSampleR * inputSampleR;
		empower = inputSampleR * addtwo; // inputSample to the third power
		inputSampleR -= (empower / 6.0);
		empower *= addtwo; // to the fifth power
		inputSampleR += (empower / 69.0);
		empower *= addtwo; //seventh
		inputSampleR -= (empower / 2530.08);
		empower *= addtwo; //ninth
		inputSampleR += (empower / 224985.6);
		empower *= addtwo; //eleventh
		inputSampleR -= (empower / 9979200.0f);
		//this is a degenerate form of a Taylor Series to approximate sin()
		
		darkSampleL = inputSampleL;
		darkSampleR = inputSampleR;
		if (avgPos > 31) avgPos = 0;
		if (slewsing > 31) {
			post32L[avgPos] = darkSampleL; post32R[avgPos] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 32; x++) {darkSampleL += post32L[x]; darkSampleR += post32R[x];}
			darkSampleL /= 32.0; darkSampleR /= 32.0;
		} if (slewsing > 15) {
			post16L[avgPos%16] = darkSampleL; post16R[avgPos%16] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 16; x++) {darkSampleL += post16L[x]; darkSampleR += post16R[x];}
			darkSampleL /= 16.0; darkSampleR /= 16.0;
		} if (slewsing > 7) {
			post8L[avgPos%8] = darkSampleL; post8R[avgPos%8] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 8; x++) {darkSampleL += post8L[x]; darkSampleR += post8R[x];}
			darkSampleL /= 8.0; darkSampleR /= 8.0;
		} if (slewsing > 3) {
			post4L[avgPos%4] = darkSampleL; post4R[avgPos%4] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 4; x++) {darkSampleL += post4L[x]; darkSampleR += post4R[x];}
			darkSampleL /= 4.0; darkSampleR /= 4.0;
		} if (slewsing > 1) {
			post2L[avgPos%2] = darkSampleL; post2R[avgPos%2] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 2; x++) {darkSampleL += post2L[x]; darkSampleR += post2R[x];}
			darkSampleL /= 2.0; darkSampleR /= 2.0; 
		} avgPos++;
		inputSampleL = (inputSampleL*(1.0-avgSlewL)) + (darkSampleL*avgSlewL);
		inputSampleR = (inputSampleR*(1.0-avgSlewR)) + (darkSampleR*avgSlewR);
		//use the previously calculated depth of the filter
		
		//begin HeadBump
		double headBumpSampleL = 0.0;
		double headBumpSampleR = 0.0;
		if (headBumpMix > 0.0) {
			headBumpL += (inputSampleL * headBumpDrive);
			headBumpL -= (headBumpL * headBumpL * headBumpL * (0.0618/sqrt(overallscale)));
			headBumpR += (inputSampleR * headBumpDrive);
			headBumpR -= (headBumpR * headBumpR * headBumpR * (0.0618/sqrt(overallscale)));
			double headBiqSampleL = (headBumpL * hdbA[hdb_a0]) + hdbA[hdb_sL1];
			hdbA[hdb_sL1] = (headBumpL * hdbA[hdb_a1]) - (headBiqSampleL * hdbA[hdb_b1]) + hdbA[hdb_sL2];
			hdbA[hdb_sL2] = (headBumpL * hdbA[hdb_a2]) - (headBiqSampleL * hdbA[hdb_b2]);
			headBumpSampleL = (headBiqSampleL * hdbB[hdb_a0]) + hdbB[hdb_sL1];
			hdbB[hdb_sL1] = (headBiqSampleL * hdbB[hdb_a1]) - (headBumpSampleL * hdbB[hdb_b1]) + hdbB[hdb_sL2];
			hdbB[hdb_sL2] = (headBiqSampleL * hdbB[hdb_a2]) - (headBumpSampleL * hdbB[hdb_b2]);
			double headBiqSampleR = (headBumpR * hdbA[hdb_a0]) + hdbA[hdb_sR1];
			hdbA[hdb_sR1] = (headBumpR * hdbA[hdb_a1]) - (headBiqSampleR * hdbA[hdb_b1]) + hdbA[hdb_sR2];
			hdbA[hdb_sR2] = (headBumpR * hdbA[hdb_a2]) - (headBiqSampleR * hdbA[hdb_b2]);
			headBumpSampleR = (headBiqSampleR * hdbB[hdb_a0]) + hdbB[hdb_sR1];
			hdbB[hdb_sR1] = (headBiqSampleR * hdbB[hdb_a1]) - (headBumpSampleR * hdbB[hdb_b1]) + hdbB[hdb_sR2];
			hdbB[hdb_sR2] = (headBiqSampleR * hdbB[hdb_a2]) - (headBumpSampleR * hdbB[hdb_b2]);
		}
		//end HeadBump
		
		inputSampleL += (headBumpSampleL * headBumpMix);
		inputSampleR += (headBumpSampleR * headBumpMix);
		
		//Dubly decode
		iirDecL = (iirDecL * (1.0 - iirDecFreq)) + (inputSampleL * iirDecFreq);
		highPart = ((inputSampleL-iirDecL)*2.628);
		highPart += avgDecL; avgDecL = (inputSampleL-iirDecL)*1.372;
		if (highPart > 1.0) highPart = 1.0; if (highPart < -1.0) highPart = -1.0;
		dubly = fabs(highPart);
		if (dubly > 0.0) {
			double adjust = log(1.0+(255.0*dubly))/2.40823996531;
			if (adjust > 0.0) dubly /= adjust;
			compDecL = (compDecL*(1.0-iirDecFreq))+(dubly*iirDecFreq);
			inputSampleL += ((highPart*compDecL)*outlyAmount);
		} //end Dubly decode L
		iirDecR = (iirDecR * (1.0 - iirDecFreq)) + (inputSampleR * iirDecFreq);
		highPart = ((inputSampleR-iirDecR)*2.628);
		highPart += avgDecR; avgDecR = (inputSampleR-iirDecR)*1.372;
		if (highPart > 1.0) highPart = 1.0; if (highPart < -1.0) highPart = -1.0;
		dubly = fabs(highPart);
		if (dubly > 0.0) {
			double adjust = log(1.0+(255.0*dubly))/2.40823996531;
			if (adjust > 0.0) dubly /= adjust;
			compDecR = (compDecR*(1.0-iirDecFreq))+(dubly*iirDecFreq);
			inputSampleR += ((highPart*compDecR)*outlyAmount);
		} //end Dubly decode R
		
		
		if (outputGain != 1.0) {
			inputSampleL *= outputGain;
			inputSampleR *= outputGain;
		}
		
		//begin ClipOnly3 as a little, compressed chunk that can be dropped into code
		double noise = 1.0-((double(fpdL)/UINT32_MAX)*0.076);
		if (wasPosClipL == true) { //current will be over
			if (inputSampleL<lastSampleL) lastSampleL=(0.9085097*noise)+(inputSampleL*(1.0-noise));
			else lastSampleL = 0.94; //~-0.2dB to nearly match ClipOnly and ClipOnly2
		} wasPosClipL = false;
		if (inputSampleL>0.9085097) {wasPosClipL=true;inputSampleL=(0.9085097*noise)+(lastSampleL*(1.0-noise));}
		if (wasNegClipL == true) { //current will be -over
			if (inputSampleL > lastSampleL) lastSampleL=(-0.9085097*noise)+(inputSampleL*(1.0-noise));
			else lastSampleL = -0.94;
		} wasNegClipL = false;
		if (inputSampleL<-0.9085097) {wasNegClipL=true;inputSampleL=(-0.9085097*noise)+(lastSampleL*(1.0-noise));}
		slewL[spacing*2] = fabs(lastSampleL-inputSampleL);
		for (int x = spacing*2; x > 0; x--) slewL[x-1] = slewL[x];
		intermediateL[spacing] = inputSampleL; inputSampleL = lastSampleL;
		//latency is however many samples equals one 44.1k sample
		for (int x = spacing; x > 0; x--) {intermediateL[x-1] = intermediateL[x];} lastSampleL = intermediateL[0];
		if (wasPosClipL || wasNegClipL) {
			for (int x = spacing; x > 0; x--) lastSampleL += intermediateL[x];
			lastSampleL /= spacing;
		} double finalSlew = 0.0;
		for (int x = spacing*2; x >= 0; x--) if (finalSlew < slewL[x]) finalSlew = slewL[x];
		double postclip = 0.94 / (1.0+(finalSlew*1.3986013));
		if (inputSampleL > postclip) inputSampleL = postclip; if (inputSampleL < -postclip) inputSampleL = -postclip;
		
		noise = 1.0-((double(fpdR)/UINT32_MAX)*0.076);
		if (wasPosClipR == true) { //current will be over
			if (inputSampleR<lastSampleR) lastSampleR=(0.9085097*noise)+(inputSampleR*(1.0-noise));
			else lastSampleR = 0.94; //~-0.2dB to nearly match ClipOnly and ClipOnly2
		} wasPosClipR = false;
		if (inputSampleR>0.9085097) {wasPosClipR=true;inputSampleR=(0.9085097*noise)+(lastSampleR*(1.0-noise));}
		if (wasNegClipR == true) { //current will be -over
			if (inputSampleR > lastSampleR) lastSampleR=(-0.9085097*noise)+(inputSampleR*(1.0-noise));
			else lastSampleR = -0.94;
		} wasNegClipR = false;
		if (inputSampleR<-0.9085097) {wasNegClipR=true;inputSampleR=(-0.9085097*noise)+(lastSampleR*(1.0-noise));}
		slewR[spacing*2] = fabs(lastSampleR-inputSampleR);
		for (int x = spacing*2; x > 0; x--) slewR[x-1] = slewR[x];
		intermediateR[spacing] = inputSampleR; inputSampleR = lastSampleR;
		//latency is however many samples equals one 44.1k sample
		for (int x = spacing; x > 0; x--) {intermediateR[x-1] = intermediateR[x];} lastSampleR = intermediateR[0];
		if (wasPosClipR || wasNegClipR) {
			for (int x = spacing; x > 0; x--) lastSampleR += intermediateR[x];
			lastSampleR /= spacing;
		} finalSlew = 0.0;
		for (int x = spacing*2; x >= 0; x--) if (finalSlew < slewR[x]) finalSlew = slewR[x];
		postclip = 0.94 / (1.0+(finalSlew*1.3986013));
		if (inputSampleR > postclip) inputSampleR = postclip; if (inputSampleR < -postclip) inputSampleR = -postclip;
		//end ClipOnly3 as a little, compressed chunk that can be dropped into code
		
		//begin 32 bit stereo floating point dither
		int expon; frexpf((float)inputSampleL, &expon);
		fpdL ^= fpdL << 13; fpdL ^= fpdL >> 17; fpdL ^= fpdL << 5;
		inputSampleL += ((double(fpdL)-uint32_t(0x7fffffff)) * 5.5e-36l * pow(2,expon+62));
		frexpf((float)inputSampleR, &expon);
		fpdR ^= fpdR << 13; fpdR ^= fpdR >> 17; fpdR ^= fpdR << 5;
		inputSampleR += ((double(fpdR)-uint32_t(0x7fffffff)) * 5.5e-36l * pow(2,expon+62));
		//end 32 bit stereo floating point dither
		
		*out1 = inputSampleL;
		*out2 = inputSampleR;

		in1++;
		in2++;
		out1++;
		out2++;
    }
}

void ToTape9::processDoubleReplacing(double **inputs, double **outputs, VstInt32 sampleFrames) 
{
    double* in1  =  inputs[0];
    double* in2  =  inputs[1];
    double* out1 = outputs[0];
    double* out2 = outputs[1];

	double overallscale = 1.0;
	overallscale /= 44100.0;
	overallscale *= getSampleRate();
	
	int spacing = floor(overallscale); //should give us working basic scaling, usually 2 or 4
	if (spacing < 1) spacing = 1; if (spacing > 16) spacing = 16;
	int slewsing = floor(overallscale*2.0);
	if (slewsing < 2) slewsing = 2; if (slewsing > 32) slewsing = 32;
	
	double inputGain = pow(A*2.0,2.0);
	
	double dublyAmount = B*2.0;
	double outlyAmount = (1.0-B)*-2.0;
	if (outlyAmount < -1.0) outlyAmount = -1.0;
	double iirEncFreq = (1.0-C)/overallscale;
	double iirDecFreq = C/overallscale;
	
	double flutDepth = pow(D,6)*overallscale*50;
	if (flutDepth > 498.0) flutDepth = 498.0;
	double flutFrequency = (0.02*pow(E,3))/overallscale;
	double bias = (F*2.0)-1.0;
	double underBias = (pow(bias,4)*0.25)/overallscale;
	double overBias = pow(1.0-bias,3)/overallscale;
	if (bias > 0.0) underBias = 0.0;
	if (bias < 0.0) overBias = 1.0/overallscale;
	
	gslew[threshold9] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold8] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold7] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold6] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold5] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold4] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold3] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold2] = overBias;
	overBias *= 1.618033988749894848204586;
	gslew[threshold1] = overBias;
	overBias *= 1.618033988749894848204586;
	
	double headBumpDrive = (G*0.1)/overallscale;
	double headBumpMix = G*0.5;
	
	hdbA[hdb_freq] = (((H*H)*175.0)+25.0)/getSampleRate();
	hdbB[hdb_freq] = hdbA[hdb_freq]*0.9375;
	hdbB[hdb_reso] = hdbA[hdb_reso] = 0.618033988749894848204586;
	hdbB[hdb_a1] = hdbA[hdb_a1] = 0.0;
	
	double K = tan(M_PI * hdbA[hdb_freq]);
	double norm = 1.0 / (1.0 + K / hdbA[hdb_reso] + K * K);
	hdbA[hdb_a0] = K / hdbA[hdb_reso] * norm;
	hdbA[hdb_a2] = -hdbA[hdb_a0];
	hdbA[hdb_b1] = 2.0 * (K * K - 1.0) * norm;
	hdbA[hdb_b2] = (1.0 - K / hdbA[hdb_reso] + K * K) * norm;
	K = tan(M_PI * hdbB[hdb_freq]);
	norm = 1.0 / (1.0 + K / hdbB[hdb_reso] + K * K);
	hdbB[hdb_a0] = K / hdbB[hdb_reso] * norm;
	hdbB[hdb_a2] = -hdbB[hdb_a0];
	hdbB[hdb_b1] = 2.0 * (K * K - 1.0) * norm;
	hdbB[hdb_b2] = (1.0 - K / hdbB[hdb_reso] + K * K) * norm;
	
	double outputGain = I*2.0;
	
    while (--sampleFrames >= 0)
    {
		double inputSampleL = *in1;
		double inputSampleR = *in2;
		if (fabs(inputSampleL)<1.18e-23) inputSampleL = fpdL * 1.18e-17;
		if (fabs(inputSampleR)<1.18e-23) inputSampleR = fpdR * 1.18e-17;
		
		if (inputGain != 1.0) {
			inputSampleL *= inputGain;
			inputSampleR *= inputGain;
		}
		
		//Dubly encode
		iirEncL = (iirEncL * (1.0 - iirEncFreq)) + (inputSampleL * iirEncFreq);
		double highPart = ((inputSampleL-iirEncL)*2.848);
		highPart += avgEncL; avgEncL = (inputSampleL-iirEncL)*1.152;
		if (highPart > 1.0) highPart = 1.0; if (highPart < -1.0) highPart = -1.0;
		double dubly = fabs(highPart);
		if (dubly > 0.0) {
			double adjust = log(1.0+(255.0*dubly))/2.40823996531;
			if (adjust > 0.0) dubly /= adjust;
			compEncL = (compEncL*(1.0-iirEncFreq))+(dubly*iirEncFreq);
			inputSampleL += ((highPart*compEncL)*dublyAmount);
		} //end Dubly encode L
		iirEncR = (iirEncR * (1.0 - iirEncFreq)) + (inputSampleR * iirEncFreq);
		highPart = ((inputSampleR-iirEncR)*2.848);
		highPart += avgEncR; avgEncR = (inputSampleR-iirEncR)*1.152;
		if (highPart > 1.0) highPart = 1.0; if (highPart < -1.0) highPart = -1.0;
		dubly = fabs(highPart);
		if (dubly > 0.0) {
			double adjust = log(1.0+(255.0*dubly))/2.40823996531;
			if (adjust > 0.0) dubly /= adjust;
			compEncR = (compEncR*(1.0-iirEncFreq))+(dubly*iirEncFreq);
			inputSampleR += ((highPart*compEncR)*dublyAmount);
		} //end Dubly encode R
		
		//begin Flutter
		if (flutDepth > 0.0) {
			if (gcount < 0 || gcount > 999) gcount = 999;
			dL[gcount] = inputSampleL;
			int count = gcount;
			double offset = flutDepth + (flutDepth * sin(sweepL));
			sweepL += nextmaxL * flutFrequency;
			if (sweepL > (M_PI*2.0)) {
				sweepL -= M_PI*2.0;
				double flutA = 0.24 + (fpdL / (double)UINT32_MAX * 0.74);
				fpdL ^= fpdL << 13; fpdL ^= fpdL >> 17; fpdL ^= fpdL << 5;
				double flutB = 0.24 + (fpdL / (double)UINT32_MAX * 0.74);
				if (fabs(flutA-sin(sweepR+nextmaxR))<fabs(flutB-sin(sweepR+nextmaxR))) nextmaxL = flutA; else nextmaxL = flutB;
			}
			count += (int)floor(offset);
			inputSampleL = (dL[count-((count > 999)?1000:0)] * (1-(offset-floor(offset))));
			inputSampleL += (dL[count+1-((count+1 > 999)?1000:0)] * (offset-floor(offset)));
			dR[gcount] = inputSampleR;
			count = gcount;
			offset = flutDepth + (flutDepth * sin(sweepR));
			sweepR += nextmaxR * flutFrequency;
			if (sweepR > (M_PI*2.0)) {
				sweepR -= M_PI*2.0;
				double flutA = 0.24 + (fpdR / (double)UINT32_MAX * 0.74);
				fpdR ^= fpdR << 13; fpdR ^= fpdR >> 17; fpdR ^= fpdR << 5;
				double flutB = 0.24 + (fpdR / (double)UINT32_MAX * 0.74);
				if (fabs(flutA-sin(sweepL+nextmaxL))<fabs(flutB-sin(sweepL+nextmaxL))) nextmaxR = flutA; else nextmaxR = flutB;
			}
			count += (int)floor(offset);
			inputSampleR = (dR[count-((count > 999)?1000:0)] * (1-(offset-floor(offset))));
			inputSampleR += (dR[count+1-((count+1 > 999)?1000:0)] * (offset-floor(offset)));
			gcount--;
		}
		//end Flutter
		
		//start bias routine
		if (fabs(bias) > 0.001) {
			for (int x = 0; x < gslew_total; x += 3) {
				if (underBias > 0.0) {
					double stuck = fabs(inputSampleL - (gslew[x]/0.975)) / underBias;
					if (stuck < 1.0) inputSampleL = (inputSampleL * stuck) + ((gslew[x]/0.975)*(1.0-stuck));
					stuck =  fabs(inputSampleR - (gslew[x+1]/0.975)) / underBias;
					if (stuck < 1.0) inputSampleR = (inputSampleR * stuck) + ((gslew[x+1]/0.975)*(1.0-stuck));
				}
				if ((inputSampleL - gslew[x]) > gslew[x+2]) inputSampleL = gslew[x] + gslew[x+2];
				if (-(inputSampleL - gslew[x]) > gslew[x+2]) inputSampleL = gslew[x] - gslew[x+2];
				gslew[x] = inputSampleL * 0.975;
				if ((inputSampleR - gslew[x+1]) > gslew[x+2]) inputSampleR = gslew[x+1] + gslew[x+2];
				if (-(inputSampleR - gslew[x+1]) > gslew[x+2]) inputSampleR = gslew[x+1] - gslew[x+2];
				gslew[x+1] = inputSampleR * 0.975;
			}
		}
		//end bias routine
		
		//begin tiny hysteresis
		double applyHysteresis = (1.0-fabs(inputSampleL))*(1.0-fabs(inputSampleL))*0.012;
		hysteresisL = fmax(fmin(hysteresisL+((inputSampleL*fabs(inputSampleL))),0.011449),-0.011449)*0.999;
		inputSampleL += (hysteresisL*applyHysteresis);
		applyHysteresis = (1.0-fabs(inputSampleR))*(1.0-fabs(inputSampleR))*0.012;
		hysteresisR = fmax(fmin(hysteresisR+((inputSampleR*fabs(inputSampleR))),0.011449),-0.011449)*0.999;
		inputSampleR += (hysteresisR*applyHysteresis);
		
		//begin TapeHack2
		double darkSampleL = inputSampleL;
		double darkSampleR = inputSampleR;
		if (avgPos > 31) avgPos = 0;
		if (slewsing > 31) {
			avg32L[avgPos] = darkSampleL; avg32R[avgPos] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 32; x++) {darkSampleL += avg32L[x]; darkSampleR += avg32R[x];}
			darkSampleL /= 32.0; darkSampleR /= 32.0;
		} if (slewsing > 15) {
			avg16L[avgPos%16] = darkSampleL; avg16R[avgPos%16] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 16; x++) {darkSampleL += avg16L[x]; darkSampleR += avg16R[x];}
			darkSampleL /= 16.0; darkSampleR /= 16.0;
		} if (slewsing > 7) {
			avg8L[avgPos%8] = darkSampleL; avg8R[avgPos%8] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 8; x++) {darkSampleL += avg8L[x]; darkSampleR += avg8R[x];}
			darkSampleL /= 8.0; darkSampleR /= 8.0;
		} if (slewsing > 3) {
			avg4L[avgPos%4] = darkSampleL; avg4R[avgPos%4] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 4; x++) {darkSampleL += avg4L[x]; darkSampleR += avg4R[x];}
			darkSampleL /= 4.0; darkSampleR /= 4.0;
		} if (slewsing > 1) {
			avg2L[avgPos%2] = darkSampleL; avg2R[avgPos%2] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 2; x++) {darkSampleL += avg2L[x]; darkSampleR += avg2R[x];}
			darkSampleL /= 2.0; darkSampleR /= 2.0; 
		} //only update avgPos after the post-distortion filter stage
		double avgSlewL = fmin(fabs(lastDarkL-inputSampleL)*0.12*overallscale,1.0);
		avgSlewL = 1.0-(1.0-avgSlewL*1.0-avgSlewL);
		inputSampleL = (inputSampleL*(1.0-avgSlewL)) + (darkSampleL*avgSlewL);
		lastDarkL = darkSampleL;
		double avgSlewR = fmin(fabs(lastDarkR-inputSampleR)*0.12*overallscale,1.0);
		avgSlewR = 1.0-(1.0-avgSlewR*1.0-avgSlewR);
		inputSampleR = (inputSampleR*(1.0-avgSlewR)) + (darkSampleR*avgSlewR);
		lastDarkR = darkSampleR;
		
		//begin TapeHack
		inputSampleL = fmax(fmin(inputSampleL,2.305929007734908),-2.305929007734908);
		double addtwo = inputSampleL * inputSampleL;
		double empower = inputSampleL * addtwo; // inputSample to the third power
		inputSampleL -= (empower / 6.0);
		empower *= addtwo; // to the fifth power
		inputSampleL += (empower / 69.0);
		empower *= addtwo; //seventh
		inputSampleL -= (empower / 2530.08);
		empower *= addtwo; //ninth
		inputSampleL += (empower / 224985.6);
		empower *= addtwo; //eleventh
		inputSampleL -= (empower / 9979200.0f);
		//this is a degenerate form of a Taylor Series to approximate sin()
		
		inputSampleR = fmax(fmin(inputSampleR,2.305929007734908),-2.305929007734908);
		addtwo = inputSampleR * inputSampleR;
		empower = inputSampleR * addtwo; // inputSample to the third power
		inputSampleR -= (empower / 6.0);
		empower *= addtwo; // to the fifth power
		inputSampleR += (empower / 69.0);
		empower *= addtwo; //seventh
		inputSampleR -= (empower / 2530.08);
		empower *= addtwo; //ninth
		inputSampleR += (empower / 224985.6);
		empower *= addtwo; //eleventh
		inputSampleR -= (empower / 9979200.0f);
		//this is a degenerate form of a Taylor Series to approximate sin()
		
		darkSampleL = inputSampleL;
		darkSampleR = inputSampleR;
		if (avgPos > 31) avgPos = 0;
		if (slewsing > 31) {
			post32L[avgPos] = darkSampleL; post32R[avgPos] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 32; x++) {darkSampleL += post32L[x]; darkSampleR += post32R[x];}
			darkSampleL /= 32.0; darkSampleR /= 32.0;
		} if (slewsing > 15) {
			post16L[avgPos%16] = darkSampleL; post16R[avgPos%16] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 16; x++) {darkSampleL += post16L[x]; darkSampleR += post16R[x];}
			darkSampleL /= 16.0; darkSampleR /= 16.0;
		} if (slewsing > 7) {
			post8L[avgPos%8] = darkSampleL; post8R[avgPos%8] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 8; x++) {darkSampleL += post8L[x]; darkSampleR += post8R[x];}
			darkSampleL /= 8.0; darkSampleR /= 8.0;
		} if (slewsing > 3) {
			post4L[avgPos%4] = darkSampleL; post4R[avgPos%4] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 4; x++) {darkSampleL += post4L[x]; darkSampleR += post4R[x];}
			darkSampleL /= 4.0; darkSampleR /= 4.0;
		} if (slewsing > 1) {
			post2L[avgPos%2] = darkSampleL; post2R[avgPos%2] = darkSampleR;
			darkSampleL = 0.0; darkSampleR = 0.0;
			for (int x = 0; x < 2; x++) {darkSampleL += post2L[x]; darkSampleR += post2R[x];}
			darkSampleL /= 2.0; darkSampleR /= 2.0; 
		} avgPos++;
		inputSampleL = (inputSampleL*(1.0-avgSlewL)) + (darkSampleL*avgSlewL);
		inputSampleR = (inputSampleR*(1.0-avgSlewR)) + (darkSampleR*avgSlewR);
		//use the previously calculated depth of the filter
		
		//begin HeadBump
		double headBumpSampleL = 0.0;
		double headBumpSampleR = 0.0;
		if (headBumpMix > 0.0) {
			headBumpL += (inputSampleL * headBumpDrive);
			headBumpL -= (headBumpL * headBumpL * headBumpL * (0.0618/sqrt(overallscale)));
			headBumpR += (inputSampleR * headBumpDrive);
			headBumpR -= (headBumpR * headBumpR * headBumpR * (0.0618/sqrt(overallscale)));
			double headBiqSampleL = (headBumpL * hdbA[hdb_a0]) + hdbA[hdb_sL1];
			hdbA[hdb_sL1] = (headBumpL * hdbA[hdb_a1]) - (headBiqSampleL * hdbA[hdb_b1]) + hdbA[hdb_sL2];
			hdbA[hdb_sL2] = (headBumpL * hdbA[hdb_a2]) - (headBiqSampleL * hdbA[hdb_b2]);
			headBumpSampleL = (headBiqSampleL * hdbB[hdb_a0]) + hdbB[hdb_sL1];
			hdbB[hdb_sL1] = (headBiqSampleL * hdbB[hdb_a1]) - (headBumpSampleL * hdbB[hdb_b1]) + hdbB[hdb_sL2];
			hdbB[hdb_sL2] = (headBiqSampleL * hdbB[hdb_a2]) - (headBumpSampleL * hdbB[hdb_b2]);
			double headBiqSampleR = (headBumpR * hdbA[hdb_a0]) + hdbA[hdb_sR1];
			hdbA[hdb_sR1] = (headBumpR * hdbA[hdb_a1]) - (headBiqSampleR * hdbA[hdb_b1]) + hdbA[hdb_sR2];
			hdbA[hdb_sR2] = (headBumpR * hdbA[hdb_a2]) - (headBiqSampleR * hdbA[hdb_b2]);
			headBumpSampleR = (headBiqSampleR * hdbB[hdb_a0]) + hdbB[hdb_sR1];
			hdbB[hdb_sR1] = (headBiqSampleR * hdbB[hdb_a1]) - (headBumpSampleR * hdbB[hdb_b1]) + hdbB[hdb_sR2];
			hdbB[hdb_sR2] = (headBiqSampleR * hdbB[hdb_a2]) - (headBumpSampleR * hdbB[hdb_b2]);
		}
		//end HeadBump
		
		inputSampleL += (headBumpSampleL * headBumpMix);
		inputSampleR += (headBumpSampleR * headBumpMix);
		
		//Dubly decode
		iirDecL = (iirDecL * (1.0 - iirDecFreq)) + (inputSampleL * iirDecFreq);
		highPart = ((inputSampleL-iirDecL)*2.628);
		highPart += avgDecL; avgDecL = (inputSampleL-iirDecL)*1.372;
		if (highPart > 1.0) highPart = 1.0; if (highPart < -1.0) highPart = -1.0;
		dubly = fabs(highPart);
		if (dubly > 0.0) {
			double adjust = log(1.0+(255.0*dubly))/2.40823996531;
			if (adjust > 0.0) dubly /= adjust;
			compDecL = (compDecL*(1.0-iirDecFreq))+(dubly*iirDecFreq);
			inputSampleL += ((highPart*compDecL)*outlyAmount);
		} //end Dubly decode L
		iirDecR = (iirDecR * (1.0 - iirDecFreq)) + (inputSampleR * iirDecFreq);
		highPart = ((inputSampleR-iirDecR)*2.628);
		highPart += avgDecR; avgDecR = (inputSampleR-iirDecR)*1.372;
		if (highPart > 1.0) highPart = 1.0; if (highPart < -1.0) highPart = -1.0;
		dubly = fabs(highPart);
		if (dubly > 0.0) {
			double adjust = log(1.0+(255.0*dubly))/2.40823996531;
			if (adjust > 0.0) dubly /= adjust;
			compDecR = (compDecR*(1.0-iirDecFreq))+(dubly*iirDecFreq);
			inputSampleR += ((highPart*compDecR)*outlyAmount);
		} //end Dubly decode R
		
		
		if (outputGain != 1.0) {
			inputSampleL *= outputGain;
			inputSampleR *= outputGain;
		}
		
		//begin ClipOnly3 as a little, compressed chunk that can be dropped into code
		double noise = 1.0-((double(fpdL)/UINT32_MAX)*0.076);
		if (wasPosClipL == true) { //current will be over
			if (inputSampleL<lastSampleL) lastSampleL=(0.9085097*noise)+(inputSampleL*(1.0-noise));
			else lastSampleL = 0.94; //~-0.2dB to nearly match ClipOnly and ClipOnly2
		} wasPosClipL = false;
		if (inputSampleL>0.9085097) {wasPosClipL=true;inputSampleL=(0.9085097*noise)+(lastSampleL*(1.0-noise));}
		if (wasNegClipL == true) { //current will be -over
			if (inputSampleL > lastSampleL) lastSampleL=(-0.9085097*noise)+(inputSampleL*(1.0-noise));
			else lastSampleL = -0.94;
		} wasNegClipL = false;
		if (inputSampleL<-0.9085097) {wasNegClipL=true;inputSampleL=(-0.9085097*noise)+(lastSampleL*(1.0-noise));}
		slewL[spacing*2] = fabs(lastSampleL-inputSampleL);
		for (int x = spacing*2; x > 0; x--) slewL[x-1] = slewL[x];
		intermediateL[spacing] = inputSampleL; inputSampleL = lastSampleL;
		//latency is however many samples equals one 44.1k sample
		for (int x = spacing; x > 0; x--) {intermediateL[x-1] = intermediateL[x];} lastSampleL = intermediateL[0];
		if (wasPosClipL || wasNegClipL) {
			for (int x = spacing; x > 0; x--) lastSampleL += intermediateL[x];
			lastSampleL /= spacing;
		} double finalSlew = 0.0;
		for (int x = spacing*2; x >= 0; x--) if (finalSlew < slewL[x]) finalSlew = slewL[x];
		double postclip = 0.94 / (1.0+(finalSlew*1.3986013));
		if (inputSampleL > postclip) inputSampleL = postclip; if (inputSampleL < -postclip) inputSampleL = -postclip;
		
		noise = 1.0-((double(fpdR)/UINT32_MAX)*0.076);
		if (wasPosClipR == true) { //current will be over
			if (inputSampleR<lastSampleR) lastSampleR=(0.9085097*noise)+(inputSampleR*(1.0-noise));
			else lastSampleR = 0.94; //~-0.2dB to nearly match ClipOnly and ClipOnly2
		} wasPosClipR = false;
		if (inputSampleR>0.9085097) {wasPosClipR=true;inputSampleR=(0.9085097*noise)+(lastSampleR*(1.0-noise));}
		if (wasNegClipR == true) { //current will be -over
			if (inputSampleR > lastSampleR) lastSampleR=(-0.9085097*noise)+(inputSampleR*(1.0-noise));
			else lastSampleR = -0.94;
		} wasNegClipR = false;
		if (inputSampleR<-0.9085097) {wasNegClipR=true;inputSampleR=(-0.9085097*noise)+(lastSampleR*(1.0-noise));}
		slewR[spacing*2] = fabs(lastSampleR-inputSampleR);
		for (int x = spacing*2; x > 0; x--) slewR[x-1] = slewR[x];
		intermediateR[spacing] = inputSampleR; inputSampleR = lastSampleR;
		//latency is however many samples equals one 44.1k sample
		for (int x = spacing; x > 0; x--) {intermediateR[x-1] = intermediateR[x];} lastSampleR = intermediateR[0];
		if (wasPosClipR || wasNegClipR) {
			for (int x = spacing; x > 0; x--) lastSampleR += intermediateR[x];
			lastSampleR /= spacing;
		} finalSlew = 0.0;
		for (int x = spacing*2; x >= 0; x--) if (finalSlew < slewR[x]) finalSlew = slewR[x];
		postclip = 0.94 / (1.0+(finalSlew*1.3986013));
		if (inputSampleR > postclip) inputSampleR = postclip; if (inputSampleR < -postclip) inputSampleR = -postclip;
		//end ClipOnly3 as a little, compressed chunk that can be dropped into code
		
		//begin 64 bit stereo floating point dither
		//int expon; frexp((double)inputSampleL, &expon);
		fpdL ^= fpdL << 13; fpdL ^= fpdL >> 17; fpdL ^= fpdL << 5;
		//inputSampleL += ((double(fpdL)-uint32_t(0x7fffffff)) * 1.1e-44l * pow(2,expon+62));
		//frexp((double)inputSampleR, &expon);
		fpdR ^= fpdR << 13; fpdR ^= fpdR >> 17; fpdR ^= fpdR << 5;
		//inputSampleR += ((double(fpdR)-uint32_t(0x7fffffff)) * 1.1e-44l * pow(2,expon+62));
		//end 64 bit stereo floating point dither
		
		*out1 = inputSampleL;
		*out2 = inputSampleR;

		in1++;
		in2++;
		out1++;
		out2++;
    }
}