diff --git a/cmake/SfizzSIMDSourceFiles.cmake b/cmake/SfizzSIMDSourceFiles.cmake
index edca20e6..efd86bb7 100644
--- a/cmake/SfizzSIMDSourceFiles.cmake
+++ b/cmake/SfizzSIMDSourceFiles.cmake
@@ -13,6 +13,7 @@ macro(sfizz_add_simd_sources SOURCES_VAR PREFIX)
             set_source_files_properties(
                 ${PREFIX}/sfizz/effects/impl/ResonantStringAVX.cpp
                 ${PREFIX}/sfizz/effects/impl/ResonantArrayAVX.cpp
+                ${PREFIX}/sfizz/SIMDHelpers.cpp
                 PROPERTIES COMPILE_FLAGS "-mavx")
         endif()
     endif()
diff --git a/src/sfizz/SIMDHelpers.cpp b/src/sfizz/SIMDHelpers.cpp
index 40a40e36..f495127f 100644
--- a/src/sfizz/SIMDHelpers.cpp
+++ b/src/sfizz/SIMDHelpers.cpp
@@ -5,7 +5,7 @@
 #include "SIMDConfig.h"
 
 #if SFIZZ_HAVE_SSE2
-#include <xmmintrin.h>
+#include <immintrin.h>
 #include <emmintrin.h>
 #endif
 
@@ -149,7 +149,17 @@ void applyGain<float>(float gain, const float* input, float* output, unsigned si
 
     if (getSIMDOpStatus(SIMDOps::gain)) {
 #if SFIZZ_CPU_FAMILY_X86_64 || SFIZZ_CPU_FAMILY_I386
-        if (cpuInfo.has_sse()) {
+        if (cpuInfo.has_avx()) {
+            const auto* lastAligned = prevAligned<32>(sentinel);
+            const auto mmGain = _mm256_set1_ps(gain);
+            while (unaligned<32>(input, output) && output < lastAligned)
+                *output++ = gain * (*input++);
+
+            while (output < lastAligned) {
+                _mm256_store_ps(output, _mm256_mul_ps(mmGain, _mm256_load_ps(input)));
+                incrementAll<8>(input, output);
+            }
+        } else if (cpuInfo.has_sse()) {
             const auto* lastAligned = prevAligned(sentinel);
             const auto mmGain = _mm_set1_ps(gain);
             while (unaligned(input, output) && output < lastAligned)
@@ -175,7 +185,16 @@ void applyGain<float>(const float* gain, const float* input, float* output, unsi
 
     if (getSIMDOpStatus(SIMDOps::gain)) {
 #if SFIZZ_CPU_FAMILY_X86_64 || SFIZZ_CPU_FAMILY_I386
-        if (cpuInfo.has_sse()) {
+        if (cpuInfo.has_avx()) {
+            const auto* lastAligned = prevAligned<32>(sentinel);
+            while (unaligned<32>(input, output) && output < lastAligned)
+                *output++ = (*gain++) * (*input++);
+
+            while (output < lastAligned) {
+                _mm256_store_ps(output, _mm256_mul_ps(_mm256_load_ps(gain), _mm256_load_ps(input)));
+                incrementAll<8>(input, output);
+            }
+        } else if (cpuInfo.has_sse()) {
             const auto* lastAligned = prevAligned(sentinel);
 
             while (unaligned(input, output) && output < lastAligned)
diff --git a/tests/SIMDHelpersT.cpp b/tests/SIMDHelpersT.cpp
index 0e15fa15..415406fd 100644
--- a/tests/SIMDHelpersT.cpp
+++ b/tests/SIMDHelpersT.cpp
@@ -273,82 +273,98 @@ TEST_CASE("[Helpers] Interleaved write SIMD vs Scalar")
 
 TEST_CASE("[Helpers] Gain, single")
 {
-    std::array<float, 5> input { 1.0f, 1.0f, 1.0f, 1.0f, 1.0f };
-    std::array<float, 5> output { 0.0f, 0.0f, 0.0f, 0.0f, 0.0f };
-    std::array<float, 5> expected { fillValue, fillValue, fillValue, fillValue, fillValue };
-    sfz::setSIMDOpStatus(sfz::SIMDOps::gain, false);
-    sfz::applyGain<float>(fillValue, input, absl::MakeSpan(output));
-    REQUIRE(output == expected);
+    std::array<float, 65> input;
+    std::array<float, 65> expected;
+    absl::c_fill(input, 1.0f);
+    absl::c_fill(expected, fillValue);
+
+    SECTION("Scalar")
+    {
+        std::array<float, 65> output;
+        sfz::setSIMDOpStatus(sfz::SIMDOps::gain, false);
+        sfz::applyGain<float>(fillValue, input, absl::MakeSpan(output));
+        REQUIRE(output == expected);
+    }
+
+    SECTION("SIMD")
+    {
+        std::array<float, 65> output;
+        sfz::setSIMDOpStatus(sfz::SIMDOps::gain, true);
+        sfz::applyGain<float>(fillValue, input, absl::MakeSpan(output));
+        REQUIRE(output == expected);
+    }
 }
 
 TEST_CASE("[Helpers] Gain, single and inplace")
 {
-    std::array<float, 5> buffer { 1.0f, 1.0f, 1.0f, 1.0f, 1.0f };
-    std::array<float, 5> expected { fillValue, fillValue, fillValue, fillValue, fillValue };
-    sfz::setSIMDOpStatus(sfz::SIMDOps::gain, false);
-    sfz::applyGain<float>(fillValue, buffer, absl::MakeSpan(buffer));
-    REQUIRE(buffer == expected);
+    std::array<float, 65> expected;
+    std::array<float, 65> buffer;
+    absl::c_fill(expected, fillValue);
+    SECTION("Scalar")
+    {
+        absl::c_fill(buffer, 1.0f);
+        sfz::setSIMDOpStatus(sfz::SIMDOps::gain, false);
+        sfz::applyGain<float>(fillValue, buffer, absl::MakeSpan(buffer));
+        REQUIRE(buffer == expected);
+    }
+    SECTION("SIMD")
+    {
+        absl::c_fill(buffer, 1.0f);
+        sfz::setSIMDOpStatus(sfz::SIMDOps::gain, false);
+        sfz::applyGain<float>(fillValue, buffer, absl::MakeSpan(buffer));
+        REQUIRE(buffer == expected);
+    }
 }
 
 TEST_CASE("[Helpers] Gain, spans")
 {
-    std::array<float, 5> input { 1.0f, 1.0f, 1.0f, 1.0f, 1.0f };
-    std::array<float, 5> gain { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f };
-    std::array<float, 5> output { 0.0f, 0.0f, 0.0f, 0.0f, 0.0f };
-    std::array<float, 5> expected { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f };
-    sfz::setSIMDOpStatus(sfz::SIMDOps::gain, false);
-    sfz::applyGain<float>(gain, input, absl::MakeSpan(output));
-    REQUIRE(output == expected);
+    std::array<float, 65> input;
+    std::array<float, 65> gain;
+    std::array<float, 65> expected;
+    absl::c_fill(input, 1.0f);
+    absl::c_iota(gain, 1.0f);
+    absl::c_iota(expected, 1.0f);
+
+    SECTION("Scalar")
+    {
+        std::array<float, 65> output;
+        sfz::setSIMDOpStatus(sfz::SIMDOps::gain, false);
+        sfz::applyGain<float>(gain, input, absl::MakeSpan(output));
+        REQUIRE(output == expected);
+    }
+
+    SECTION("SIMD")
+    {
+        std::array<float, 65> output;
+        sfz::setSIMDOpStatus(sfz::SIMDOps::gain, true);
+        sfz::applyGain<float>(gain, input, absl::MakeSpan(output));
+        REQUIRE(output == expected);
+    }
 }
 
 TEST_CASE("[Helpers] Gain, spans and inplace")
 {
-    std::array<float, 5> buffer { 1.0f, 1.0f, 1.0f, 1.0f, 1.0f };
-    std::array<float, 5> gain { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f };
-    std::array<float, 5> expected { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f };
-    sfz::setSIMDOpStatus(sfz::SIMDOps::gain, false);
-    sfz::applyGain<float>(gain, buffer, absl::MakeSpan(buffer));
-    REQUIRE(buffer == expected);
-}
+    std::array<float, 65> buffer;
+    std::array<float, 65> gain;
+    std::array<float, 65> expected;
+    absl::c_iota(gain, 1.0f);
+    absl::c_iota(expected, 1.0f);
 
-TEST_CASE("[Helpers] Gain, single (SIMD)")
-{
-    std::array<float, 5> input { 1.0f, 1.0f, 1.0f, 1.0f, 1.0f };
-    std::array<float, 5> output { 0.0f, 0.0f, 0.0f, 0.0f, 0.0f };
-    std::array<float, 5> expected { fillValue, fillValue, fillValue, fillValue, fillValue };
-    sfz::setSIMDOpStatus(sfz::SIMDOps::gain, true);
-    sfz::applyGain<float>(fillValue, input, absl::MakeSpan(output));
-    REQUIRE(output == expected);
-}
+    SECTION("Scalar")
+    {
+        absl::c_fill(buffer, 1.0f);
+        sfz::setSIMDOpStatus(sfz::SIMDOps::gain, false);
+        sfz::applyGain<float>(gain, buffer, absl::MakeSpan(buffer));
+        REQUIRE(buffer == expected);
+    }
 
-TEST_CASE("[Helpers] Gain, single and inplace (SIMD)")
-{
-    std::array<float, 5> buffer { 1.0f, 1.0f, 1.0f, 1.0f, 1.0f };
-    std::array<float, 5> expected { fillValue, fillValue, fillValue, fillValue, fillValue };
-    sfz::setSIMDOpStatus(sfz::SIMDOps::gain, true);
-    sfz::applyGain<float>(fillValue, buffer, absl::MakeSpan(buffer));
-    REQUIRE(buffer == expected);
-}
-
-TEST_CASE("[Helpers] Gain, spans (SIMD)")
-{
-    std::array<float, 5> input { 1.0f, 1.0f, 1.0f, 1.0f, 1.0f };
-    std::array<float, 5> gain { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f };
-    std::array<float, 5> output { 0.0f, 0.0f, 0.0f, 0.0f, 0.0f };
-    std::array<float, 5> expected { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f };
-    sfz::setSIMDOpStatus(sfz::SIMDOps::gain, true);
-    sfz::applyGain<float>(gain, input, absl::MakeSpan(output));
-    REQUIRE(output == expected);
-}
-
-TEST_CASE("[Helpers] Gain, spans and inplace (SIMD)")
-{
-    std::array<float, 5> buffer { 1.0f, 1.0f, 1.0f, 1.0f, 1.0f };
-    std::array<float, 5> gain { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f };
-    std::array<float, 5> expected { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f };
-    sfz::setSIMDOpStatus(sfz::SIMDOps::gain, true);
-    sfz::applyGain<float>(gain, buffer, absl::MakeSpan(buffer));
-    REQUIRE(buffer == expected);
+    SECTION("SIMD")
+    {
+        absl::c_fill(buffer, 1.0f);
+        sfz::setSIMDOpStatus(sfz::SIMDOps::gain, false);
+        sfz::applyGain<float>(gain, buffer, absl::MakeSpan(buffer));
+        REQUIRE(buffer == expected);
+    }
 }
 
 TEST_CASE("[Helpers] Linear Ramp")