diff --git a/devtools/HIIRDesigner.cpp b/devtools/HIIRDesigner.cpp
index bbb640fd..d021764a 100644
--- a/devtools/HIIRDesigner.cpp
+++ b/devtools/HIIRDesigner.cpp
@@ -199,7 +199,14 @@ static void generate_cpp_upsampler(const Stage *stages, int num_stages)
 
     printf("\t" "static int recommendedBuffer(int factor, int spl)\n");
     printf("\t" "{\n");
-    printf("\t\t" "return factor * spl;\n");
+    printf("\t\t" "switch (factor) {\n");
+    printf("\t\t" "case 2:\n");
+    printf("\t\t\t" "return 0;\n");
+    printf("\t\t" "case 4:\n");
+    printf("\t\t\t" "return 2 * spl;\n");
+    printf("\t\t" "default:\n");
+    printf("\t\t\t" "return factor * spl;\n");
+    printf("\t\t" "}\n");
     printf("\t" "}\n");
 
     printf("\t" "static bool canProcess(int factor)\n");
@@ -235,14 +242,25 @@ static void generate_cpp_upsampler(const Stage *stages, int num_stages)
     printf("\t" "}\n");
 
     for (int n = 1; n <= num_stages; ++n) {
+        // special case factor=2, buffer not required
+        if (stages[n - 1].factor == 2) {
+            printf("\t" "void process2x(const float *in, float *out, int spl, float * = nullptr, int = 0)\n");
+            printf("\t" "{\n");
+            printf("\t\t" "up2_.process_block(out, in, spl);\n");
+            printf("\t" "}\n");
+            continue;
+        }
         printf("\t" "void process%dx(const float *in, float *out, int spl, float *temp, int ntemp)\n", stages[n - 1].factor);
         printf("\t" "{\n");
-        printf("\t\t" "int maxspl = ntemp / %d;\n", stages[n - 1].factor);
-        printf("\t\t" "ASSERT(maxspl >= 0);\n");
+        // special case factor=4, only 1 buffer required
+        if (stages[n - 1].factor > 4)
+            printf("\t\t" "int maxspl = ntemp / %d;\n", stages[n - 1].factor);
+        else
+            printf("\t\t" "int maxspl = ntemp / %d;\n", stages[n - 1].factor / 2);
+        printf("\t\t" "ASSERT(maxspl > 0);\n");
         printf("\t\t" "float *t1 = temp;\n");
-        printf("\t\t" "float *t2 = temp + %d * maxspl;\n", stages[n - 1].factor / 2);
-        printf("\t\t" "(void)t1;\n");
-        printf("\t\t" "(void)t2;\n");
+        if (stages[n - 1].factor > 4)
+            printf("\t\t" "float *t2 = temp + %d * maxspl;\n", stages[n - 1].factor / 2);
         printf("\t\t" "while (spl > 0) {\n");
         printf("\t\t\t" "int curspl = (spl < maxspl) ? spl : maxspl;\n");
         for (int i = 0; i < n; ++i) {
@@ -294,7 +312,14 @@ static void generate_cpp_downsampler(const Stage *stages, int num_stages)
 
     printf("\t" "static int recommendedBuffer(int factor, int spl)\n");
     printf("\t" "{\n");
-    printf("\t\t" "return factor * spl;\n");
+    printf("\t\t" "switch (factor) {\n");
+    printf("\t\t" "case 2:\n");
+    printf("\t\t\t" "return 0;\n");
+    printf("\t\t" "case 4:\n");
+    printf("\t\t\t" "return 2 * spl;\n");
+    printf("\t\t" "default:\n");
+    printf("\t\t\t" "return factor * spl;\n");
+    printf("\t\t" "}\n");
     printf("\t" "}\n");
 
     printf("\t" "static bool canProcess(int factor)\n");
@@ -330,14 +355,25 @@ static void generate_cpp_downsampler(const Stage *stages, int num_stages)
     printf("\t" "}\n");
 
     for (int n = 1; n <= num_stages; ++n) {
+        // special case factor=2, buffer not required
+        if (stages[n - 1].factor == 2) {
+            printf("\t" "void process2x(const float *in, float *out, int spl, float * = nullptr, int = 0)\n");
+            printf("\t" "{\n");
+            printf("\t\t" "down2_.process_block(out, in, spl);\n");
+            printf("\t" "}\n");
+            continue;
+        }
         printf("\t" "void process%dx(const float *in, float *out, int spl, float *temp, int ntemp)\n", stages[n - 1].factor);
         printf("\t" "{\n");
-        printf("\t\t" "int maxspl = ntemp / %d;\n", stages[n - 1].factor);
-        printf("\t\t" "ASSERT(maxspl >= 0);\n");
+        // special case factor=4, only 1 buffer required
+        if (stages[n - 1].factor > 4)
+            printf("\t\t" "int maxspl = ntemp / %d;\n", stages[n - 1].factor);
+        else
+            printf("\t\t" "int maxspl = ntemp / %d;\n", stages[n - 1].factor / 2);
+        printf("\t\t" "ASSERT(maxspl > 0);\n");
         printf("\t\t" "float *t1 = temp;\n");
-        printf("\t\t" "float *t2 = temp + %d * maxspl;\n", stages[n - 1].factor / 2);
-        printf("\t\t" "(void)t1;\n");
-        printf("\t\t" "(void)t2;\n");
+        if (stages[n - 1].factor > 4)
+            printf("\t\t" "float *t2 = temp + %d * maxspl;\n", stages[n - 1].factor / 2);
         printf("\t\t" "while (spl > 0) {\n");
         printf("\t\t\t" "int curspl = (spl < maxspl) ? spl : maxspl;\n");
         for (int i = 0; i < n; ++i) {
diff --git a/src/sfizz/OversamplerHelpers.hxx b/src/sfizz/OversamplerHelpers.hxx
index 49cf3f60..8e1bef52 100644
--- a/src/sfizz/OversamplerHelpers.hxx
+++ b/src/sfizz/OversamplerHelpers.hxx
@@ -1,8 +1,8 @@
-// SPDX-License-Identifier: BSD-2-Clause
 
-// This code is part of the sfizz library and is licensed under a BSD 2-clause
-// license. You should have receive a LICENSE.md file along with the code.
-// If not, contact the sfizz maintainers at https://github.com/sfztools/sfizz
+
+
+
+
 
 //------------------------------------------------------------------------------
 // This is generated by the Sfizz HIIR designer
@@ -85,7 +85,14 @@ public:
 	}
 	static int recommendedBuffer(int factor, int spl)
 	{
-		return factor * spl;
+		switch (factor) {
+		case 2:
+			return 0;
+		case 4:
+			return 2 * spl;
+		default:
+			return factor * spl;
+		}
 	}
 	static bool canProcess(int factor)
 	{
@@ -135,30 +142,15 @@ public:
 			break;
 		}
 	}
-	void process2x(const float *in, float *out, int spl, float *temp, int ntemp)
+	void process2x(const float *in, float *out, int spl, float * = nullptr, int = 0)
 	{
-		int maxspl = ntemp / 2;
-		ASSERT(maxspl >= 0);
-		float *t1 = temp;
-		float *t2 = temp + 1 * maxspl;
-		(void)t1;
-		(void)t2;
-		while (spl > 0) {
-			int curspl = (spl < maxspl) ? spl : maxspl;
-			up2_.process_block(out, in, 1 * curspl);
-			in += curspl;
-			out += curspl;
-			spl -= curspl;
-		}
+		up2_.process_block(out, in, spl);
 	}
 	void process4x(const float *in, float *out, int spl, float *temp, int ntemp)
 	{
-		int maxspl = ntemp / 4;
-		ASSERT(maxspl >= 0);
+		int maxspl = ntemp / 2;
+		ASSERT(maxspl > 0);
 		float *t1 = temp;
-		float *t2 = temp + 2 * maxspl;
-		(void)t1;
-		(void)t2;
 		while (spl > 0) {
 			int curspl = (spl < maxspl) ? spl : maxspl;
 			up2_.process_block(t1, in, 1 * curspl);
@@ -171,11 +163,9 @@ public:
 	void process8x(const float *in, float *out, int spl, float *temp, int ntemp)
 	{
 		int maxspl = ntemp / 8;
-		ASSERT(maxspl >= 0);
+		ASSERT(maxspl > 0);
 		float *t1 = temp;
 		float *t2 = temp + 4 * maxspl;
-		(void)t1;
-		(void)t2;
 		while (spl > 0) {
 			int curspl = (spl < maxspl) ? spl : maxspl;
 			up2_.process_block(t1, in, 1 * curspl);
@@ -189,11 +179,9 @@ public:
 	void process16x(const float *in, float *out, int spl, float *temp, int ntemp)
 	{
 		int maxspl = ntemp / 16;
-		ASSERT(maxspl >= 0);
+		ASSERT(maxspl > 0);
 		float *t1 = temp;
 		float *t2 = temp + 8 * maxspl;
-		(void)t1;
-		(void)t2;
 		while (spl > 0) {
 			int curspl = (spl < maxspl) ? spl : maxspl;
 			up2_.process_block(t1, in, 1 * curspl);
@@ -208,11 +196,9 @@ public:
 	void process32x(const float *in, float *out, int spl, float *temp, int ntemp)
 	{
 		int maxspl = ntemp / 32;
-		ASSERT(maxspl >= 0);
+		ASSERT(maxspl > 0);
 		float *t1 = temp;
 		float *t2 = temp + 16 * maxspl;
-		(void)t1;
-		(void)t2;
 		while (spl > 0) {
 			int curspl = (spl < maxspl) ? spl : maxspl;
 			up2_.process_block(t1, in, 1 * curspl);
@@ -228,11 +214,9 @@ public:
 	void process64x(const float *in, float *out, int spl, float *temp, int ntemp)
 	{
 		int maxspl = ntemp / 64;
-		ASSERT(maxspl >= 0);
+		ASSERT(maxspl > 0);
 		float *t1 = temp;
 		float *t2 = temp + 32 * maxspl;
-		(void)t1;
-		(void)t2;
 		while (spl > 0) {
 			int curspl = (spl < maxspl) ? spl : maxspl;
 			up2_.process_block(t1, in, 1 * curspl);
@@ -249,11 +233,9 @@ public:
 	void process128x(const float *in, float *out, int spl, float *temp, int ntemp)
 	{
 		int maxspl = ntemp / 128;
-		ASSERT(maxspl >= 0);
+		ASSERT(maxspl > 0);
 		float *t1 = temp;
 		float *t2 = temp + 64 * maxspl;
-		(void)t1;
-		(void)t2;
 		while (spl > 0) {
 			int curspl = (spl < maxspl) ? spl : maxspl;
 			up2_.process_block(t1, in, 1 * curspl);
@@ -302,7 +284,14 @@ public:
 	}
 	static int recommendedBuffer(int factor, int spl)
 	{
-		return factor * spl;
+		switch (factor) {
+		case 2:
+			return 0;
+		case 4:
+			return 2 * spl;
+		default:
+			return factor * spl;
+		}
 	}
 	static bool canProcess(int factor)
 	{
@@ -352,30 +341,15 @@ public:
 			break;
 		}
 	}
-	void process2x(const float *in, float *out, int spl, float *temp, int ntemp)
+	void process2x(const float *in, float *out, int spl, float * = nullptr, int = 0)
 	{
-		int maxspl = ntemp / 2;
-		ASSERT(maxspl >= 0);
-		float *t1 = temp;
-		float *t2 = temp + 1 * maxspl;
-		(void)t1;
-		(void)t2;
-		while (spl > 0) {
-			int curspl = (spl < maxspl) ? spl : maxspl;
-			down2_.process_block(out, in, 1 * curspl);
-			in += curspl;
-			out += curspl;
-			spl -= curspl;
-		}
+		down2_.process_block(out, in, spl);
 	}
 	void process4x(const float *in, float *out, int spl, float *temp, int ntemp)
 	{
-		int maxspl = ntemp / 4;
-		ASSERT(maxspl >= 0);
+		int maxspl = ntemp / 2;
+		ASSERT(maxspl > 0);
 		float *t1 = temp;
-		float *t2 = temp + 2 * maxspl;
-		(void)t1;
-		(void)t2;
 		while (spl > 0) {
 			int curspl = (spl < maxspl) ? spl : maxspl;
 			down4_.process_block(t1, in, 2 * curspl);
@@ -388,11 +362,9 @@ public:
 	void process8x(const float *in, float *out, int spl, float *temp, int ntemp)
 	{
 		int maxspl = ntemp / 8;
-		ASSERT(maxspl >= 0);
+		ASSERT(maxspl > 0);
 		float *t1 = temp;
 		float *t2 = temp + 4 * maxspl;
-		(void)t1;
-		(void)t2;
 		while (spl > 0) {
 			int curspl = (spl < maxspl) ? spl : maxspl;
 			down8_.process_block(t1, in, 4 * curspl);
@@ -406,11 +378,9 @@ public:
 	void process16x(const float *in, float *out, int spl, float *temp, int ntemp)
 	{
 		int maxspl = ntemp / 16;
-		ASSERT(maxspl >= 0);
+		ASSERT(maxspl > 0);
 		float *t1 = temp;
 		float *t2 = temp + 8 * maxspl;
-		(void)t1;
-		(void)t2;
 		while (spl > 0) {
 			int curspl = (spl < maxspl) ? spl : maxspl;
 			down16_.process_block(t1, in, 8 * curspl);
@@ -425,11 +395,9 @@ public:
 	void process32x(const float *in, float *out, int spl, float *temp, int ntemp)
 	{
 		int maxspl = ntemp / 32;
-		ASSERT(maxspl >= 0);
+		ASSERT(maxspl > 0);
 		float *t1 = temp;
 		float *t2 = temp + 16 * maxspl;
-		(void)t1;
-		(void)t2;
 		while (spl > 0) {
 			int curspl = (spl < maxspl) ? spl : maxspl;
 			down32_.process_block(t1, in, 16 * curspl);
@@ -445,11 +413,9 @@ public:
 	void process64x(const float *in, float *out, int spl, float *temp, int ntemp)
 	{
 		int maxspl = ntemp / 64;
-		ASSERT(maxspl >= 0);
+		ASSERT(maxspl > 0);
 		float *t1 = temp;
 		float *t2 = temp + 32 * maxspl;
-		(void)t1;
-		(void)t2;
 		while (spl > 0) {
 			int curspl = (spl < maxspl) ? spl : maxspl;
 			down64_.process_block(t1, in, 32 * curspl);
@@ -466,11 +432,9 @@ public:
 	void process128x(const float *in, float *out, int spl, float *temp, int ntemp)
 	{
 		int maxspl = ntemp / 128;
-		ASSERT(maxspl >= 0);
+		ASSERT(maxspl > 0);
 		float *t1 = temp;
 		float *t2 = temp + 64 * maxspl;
-		(void)t1;
-		(void)t2;
 		while (spl > 0) {
 			int curspl = (spl < maxspl) ? spl : maxspl;
 			down128_.process_block(t1, in, 64 * curspl);