diff --git a/spa/plugins/audioconvert/resample-native-neon.c b/spa/plugins/audioconvert/resample-native-neon.c
index 97390e3d5..afe68914a 100644
--- a/spa/plugins/audioconvert/resample-native-neon.c
+++ b/spa/plugins/audioconvert/resample-native-neon.c
@@ -43,10 +43,8 @@ static void inner_product_neon(float *d, const float * SPA_RESTRICT s,
 		"      bne 4f\n"
 		"      b 5f\n"
 		"1:"
-		"      ld1 {v4.4s,  v5.4s}, [%[taps]], #32\n"
-		"      ld1 {v8.4s,  v9.4s}, [%[s]], #32\n"
-		"      ld1 {v6.4s,  v7.4s}, [%[taps]], #32\n"
-		"      ld1 {v10.4s, v11.4s}, [%[s]], #32\n"
+		"      ld1 {v4.4s,  v5.4s, v6.4s,  v7.4s}, [%[taps]], #64\n"
+		"      ld1 {v8.4s,  v9.4s, v10.4s, v11.4s}, [%[s]], #64\n"
 		"      subs %[n_taps], %[n_taps], #16\n"
 		"      fmul v0.4s, v4.4s, v8.4s\n"
 		"      fmul v1.4s, v5.4s, v9.4s\n"
@@ -54,10 +52,8 @@ static void inner_product_neon(float *d, const float * SPA_RESTRICT s,
 		"      fmul v3.4s, v7.4s, v11.4s\n"
 		"      beq 3f\n"
 		"2:"
-		"      ld1 { v4.4s,  v5.4s}, [%[taps]], #32\n"
-		"      ld1 { v8.4s,  v9.4s}, [%[s]], #32\n"
-		"      ld1 { v6.4s,  v7.4s}, [%[taps]], #32\n"
-		"      ld1 { v10.4s, v11.4s}, [%[s]], #32\n"
+		"      ld1 {v4.4s,  v5.4s, v6.4s,  v7.4s}, [%[taps]], #64\n"
+		"      ld1 {v8.4s,  v9.4s, v10.4s, v11.4s}, [%[s]], #64\n"
 		"      subs %[n_taps], %[n_taps], #16\n"
 		"      fmla v0.4s, v4.4s, v8.4s\n"
 		"      fmla v1.4s, v5.4s, v9.4s\n"
@@ -71,8 +67,8 @@ static void inner_product_neon(float *d, const float * SPA_RESTRICT s,
 		"      fadd v0.4s, v4.4s, v5.4s\n"
 		"      beq 5f\n"
 		"4:"
-		"      ld1 { v6.4s  }, [%[taps]], #16\n"
-		"      ld1 { v10.4s }, [%[s]], #16\n"
+		"      ld1 {v6.4s}, [%[taps]], #16\n"
+		"      ld1 {v10.4s}, [%[s]], #16\n"
 		"      subs %[remainder], %[remainder], #4\n"
 		"      fmla v0.4s, v6.4s, v10.4s\n"
 		"      bne 4b\n"
@@ -158,9 +154,9 @@ static void inner_product_ip_neon(float *d, const float * SPA_RESTRICT s,
 		"      fmul v3.4s, v7.4s, v9.4s\n"
 		"      beq 3f\n"
 		"2:"
-		"      ld1 { v4.4s, v5.4s}, [%[t0]], #32\n"
-		"      ld1 { v8.4s, v9.4s}, [%[s]], #32\n"
-		"      ld1 { v6.4s, v7.4s}, [%[t1]], #32\n"
+		"      ld1 {v4.4s, v5.4s}, [%[t0]], #32\n"
+		"      ld1 {v8.4s, v9.4s}, [%[s]], #32\n"
+		"      ld1 {v6.4s, v7.4s}, [%[t1]], #32\n"
 		"      subs %[n_taps], %[n_taps], #8\n"
 		"      fmla v0.4s, v4.4s, v8.4s\n"
 		"      fmla v1.4s, v5.4s, v9.4s\n"