diff --git a/libretro-common/conversion/float_to_s16_neon.c b/libretro-common/conversion/float_to_s16_neon.c
new file mode 100644
index 0000000000..4b9be62329
--- /dev/null
+++ b/libretro-common/conversion/float_to_s16_neon.c
@@ -0,0 +1,64 @@
+/* Copyright  (C) 2010-2016 The RetroArch team
+ *
+ * ---------------------------------------------------------------------------------------
+ * The following license statement only applies to this file (float_to_s16_neon.S).
+ * ---------------------------------------------------------------------------------------
+ *
+ * Permission is hereby granted, free of charge,
+ * to any person obtaining a copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation the rights to
+ * use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,
+ * INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+#if defined(__ARM_NEON__)
+
+#if defined(__thumb__)
+#define DECL_ARMMODE(x) "  .align 2\n" "  .global " x "\n" "  .thumb\n" "  .thumb_func\n" "  .type " x ", %function\n" x ":\n"
+#else
+#define DECL_ARMMODE(x) "  .align 4\n" "  .global " x "\n" "  .arm\n" x ":\n"
+#endif
+
+asm(
+    DECL_ARMMODE("convert_float_s16_asm")
+    DECL_ARMMODE("_convert_float_s16_asm")
+    "# convert_float_s16_asm(int16_t *out, const float *in, size_t samples)\n"
+    "   # Hacky way to get a constant of 2^15.\n"
+    "   # ((2^4)^2)^2 * 0.5 = 2^15\n"
+    "   vmov.f32 q8, #16.0\n"
+    "   vmov.f32 q9, #0.5\n"
+    "   vmul.f32 q8, q8, q8\n"
+    "   vmul.f32 q8, q8, q8\n"
+    "   vmul.f32 q8, q8, q9\n"
+    "\n"
+    "1:\n"
+    "   # Preload here?\n"
+    "   vld1.f32 {q0-q1}, [r1]!\n"
+    "\n"
+    "   vmul.f32 q0, q0, q8\n"
+    "   vmul.f32 q1, q1, q8\n"
+    "\n"
+    "   vcvt.s32.f32 q0, q0\n"
+    "   vcvt.s32.f32 q1, q1\n"
+    "\n"
+    "   vqmovn.s32 d4, q0\n"
+    "   vqmovn.s32 d5, q1\n"
+    "\n"
+    "   vst1.f32 {d4-d5}, [r0]!\n"
+    "\n"
+    "   # Guaranteed to get samples in multiples of 8.\n"
+    "   subs r2, r2, #8\n"
+    "   bne 1b\n"
+    "\n"
+    "   bx lr\n"
+    "\n"
+    );
+#endif
diff --git a/libretro-common/conversion/s16_to_float_neon.c b/libretro-common/conversion/s16_to_float_neon.c
new file mode 100644
index 0000000000..e6b24c00c3
--- /dev/null
+++ b/libretro-common/conversion/s16_to_float_neon.c
@@ -0,0 +1,71 @@
+/* Copyright  (C) 2010-2016 The RetroArch team
+ *
+ * ---------------------------------------------------------------------------------------
+ * The following license statement only applies to this file (s16_to_float_neon.S).
+ * ---------------------------------------------------------------------------------------
+ *
+ * Permission is hereby granted, free of charge,
+ * to any person obtaining a copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation the rights to
+ * use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,
+ * INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+#if defined(__ARM_NEON__)
+
+#if defined(__thumb__)
+#define DECL_ARMMODE(x) "  .align 2\n" "  .global " x "\n" "  .thumb\n" "  .thumb_func\n" "  .type " x ", %function\n" x ":\n"
+#else
+#define DECL_ARMMODE(x) "  .align 4\n" "  .global " x "\n" "  .arm\n" x ":\n"
+#endif
+
+asm(
+    DECL_ARMMODE("convert_s16_float_asm")
+    DECL_ARMMODE("_convert_s16_float_asm")
+    "# convert_s16_float_asm(float *out, const int16_t *in, size_t samples, const float *gain)\n"
+    "   # Hacky way to get a constant of 2^-15.\n"
+    "   # Might be faster to just load a constant from memory.\n"
+    "   # It's just done once however ...\n"
+    "   vmov.f32 q8, #0.25\n"
+    "   vmul.f32 q8, q8, q8\n"
+    "   vmul.f32 q8, q8, q8\n"
+    "   vmul.f32 q8, q8, q8\n"
+    "   vadd.f32 q8, q8, q8\n"
+    "\n"
+    "   # Apply gain\n"
+    "   vld1.f32 {d6[0]}, [r3]\n"
+    "   vmul.f32 q8, q8, d6[0]\n"
+    "\n"
+    "1:\n"
+    "   # Preload here?\n"
+    "   vld1.s16 {q0}, [r1]!\n"
+    "\n"
+    "   # Widen to 32-bit\n"
+    "   vmovl.s16 q1, d0\n"
+    "   vmovl.s16 q2, d1\n"
+    "\n"
+    "   # Convert to float\n"
+    "   vcvt.f32.s32 q1, q1\n"
+    "   vcvt.f32.s32 q2, q2\n"
+    "\n"
+    "   vmul.f32 q1, q1, q8\n"
+    "   vmul.f32 q2, q2, q8\n"
+    "\n"
+    "   vst1.f32 {q1-q2}, [r0]!\n"
+    "\n"
+    "   # Guaranteed to get samples in multiples of 8.\n"
+    "   subs r2, r2, #8\n"
+    "   bne 1b\n"
+    "\n"
+    "   bx lr\n"
+    "\n"
+    );
+#endif