diff --git a/src/core/NEON/wrapper/scalar/add.h b/src/core/NEON/wrapper/scalar/add.h
index 642d926..2ec8886 100644
--- a/src/core/NEON/wrapper/scalar/add.h
+++ b/src/core/NEON/wrapper/scalar/add.h
@@ -32,22 +32,22 @@
 {
 inline uint8_t add_sat(const uint8_t &a, const uint8_t &b)
 {
-    const uint8x8_t va = { a, 0, 0, 0, 0, 0, 0, 0 };
-    const uint8x8_t vb = { b, 0, 0, 0, 0, 0, 0, 0 };
+    const uint8x8_t va = {a, 0, 0, 0, 0, 0, 0, 0};
+    const uint8x8_t vb = {b, 0, 0, 0, 0, 0, 0, 0};
     return vget_lane_u8(vqadd_u8(va, vb), 0);
 }
 
 inline int16_t add_sat(const int16_t &a, const int16_t &b)
 {
-    const int16x4_t va = { a, 0, 0, 0 };
-    const int16x4_t vb = { b, 0, 0, 0 };
+    const int16x4_t va = {a, 0, 0, 0};
+    const int16x4_t vb = {b, 0, 0, 0};
     return vget_lane_s16(vqadd_s16(va, vb), 0);
 }
 
 inline int32_t add_sat(const int32_t &a, const int32_t &b)
 {
-    const int32x2_t va = { a, 0 };
-    const int32x2_t vb = { b, 0 };
+    const int32x2_t va = {a, 0};
+    const int32x2_t vb = {b, 0};
     return vget_lane_s32(vqadd_s32(va, vb), 0);
 }