diff --git a/test/fdct4x4_test.cc b/test/fdct4x4_test.cc
index 1270dae94..444b0209d 100644
--- a/test/fdct4x4_test.cc
+++ b/test/fdct4x4_test.cc
@@ -438,14 +438,12 @@ INSTANTIATE_TEST_CASE_P(C, Trans4x4WHT,
                                                      VPX_BITS_8)));
 #endif  // CONFIG_VP9_HIGHBITDEPTH
 
-#if HAVE_NEON_ASM && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
+#if HAVE_NEON && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(NEON, Trans4x4DCT,
                         ::testing::Values(make_tuple(&vpx_fdct4x4_c,
                                                      &vpx_idct4x4_16_add_neon,
                                                      0, VPX_BITS_8)));
-#endif  // HAVE_NEON_ASM && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
-
-#if HAVE_NEON && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
+#if !CONFIG_VP9_HIGHBITDEPTH
 INSTANTIATE_TEST_CASE_P(
     NEON, Trans4x4HT,
     ::testing::Values(
@@ -453,7 +451,8 @@ INSTANTIATE_TEST_CASE_P(
         make_tuple(&vp9_fht4x4_c, &vp9_iht4x4_16_add_neon, 1, VPX_BITS_8),
         make_tuple(&vp9_fht4x4_c, &vp9_iht4x4_16_add_neon, 2, VPX_BITS_8),
         make_tuple(&vp9_fht4x4_c, &vp9_iht4x4_16_add_neon, 3, VPX_BITS_8)));
-#endif  // HAVE_NEON && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
+#endif  // !CONFIG_VP9_HIGHBITDEPTH
+#endif  // HAVE_NEON && !CONFIG_EMULATE_HARDWARE
 
 #if HAVE_SSE2 && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
diff --git a/test/fdct8x8_test.cc b/test/fdct8x8_test.cc
index 409a76bfc..e40340490 100644
--- a/test/fdct8x8_test.cc
+++ b/test/fdct8x8_test.cc
@@ -670,14 +670,17 @@ INSTANTIATE_TEST_CASE_P(
         make_tuple(&vp9_fht8x8_c, &vp9_iht8x8_64_add_c, 3, VPX_BITS_8)));
 #endif  // CONFIG_VP9_HIGHBITDEPTH
 
-#if HAVE_NEON_ASM && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
+#if HAVE_NEON && !CONFIG_EMULATE_HARDWARE
+#if CONFIG_VP9_HIGHBITDEPTH
+INSTANTIATE_TEST_CASE_P(NEON, FwdTrans8x8DCT,
+                        ::testing::Values(make_tuple(&vpx_fdct8x8_c,
+                                                     &vpx_idct8x8_64_add_neon,
+                                                     0, VPX_BITS_8)));
+#else   // !CONFIG_VP9_HIGHBITDEPTH
 INSTANTIATE_TEST_CASE_P(NEON, FwdTrans8x8DCT,
                         ::testing::Values(make_tuple(&vpx_fdct8x8_neon,
                                                      &vpx_idct8x8_64_add_neon,
                                                      0, VPX_BITS_8)));
-#endif  // HAVE_NEON_ASM && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
-
-#if HAVE_NEON && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
     NEON, FwdTrans8x8HT,
     ::testing::Values(
@@ -685,6 +688,7 @@ INSTANTIATE_TEST_CASE_P(
         make_tuple(&vp9_fht8x8_c, &vp9_iht8x8_64_add_neon, 1, VPX_BITS_8),
         make_tuple(&vp9_fht8x8_c, &vp9_iht8x8_64_add_neon, 2, VPX_BITS_8),
         make_tuple(&vp9_fht8x8_c, &vp9_iht8x8_64_add_neon, 3, VPX_BITS_8)));
+#endif  // CONFIG_VP9_HIGHBITDEPTH
 #endif  // HAVE_NEON && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
 
 #if HAVE_SSE2 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
diff --git a/test/partial_idct_test.cc b/test/partial_idct_test.cc
index 59371f7e3..bc214ad89 100644
--- a/test/partial_idct_test.cc
+++ b/test/partial_idct_test.cc
@@ -280,10 +280,18 @@ INSTANTIATE_TEST_CASE_P(
     NEON, PartialIDctTest,
     ::testing::Values(make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
                                  &vpx_idct32x32_1_add_neon, TX_32X32, 1),
+                      make_tuple(&vpx_fdct32x32_c, &vpx_idct32x32_1024_add_c,
+                                 &vpx_idct32x32_34_add_neon, TX_32X32, 34),
                       make_tuple(&vpx_fdct16x16_c, &vpx_idct16x16_256_add_c,
                                  &vpx_idct16x16_1_add_neon, TX_16X16, 1),
+                      make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
+                                 &vpx_idct8x8_64_add_neon, TX_8X8, 64),
+                      make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
+                                 &vpx_idct8x8_12_add_neon, TX_8X8, 12),
                       make_tuple(&vpx_fdct8x8_c, &vpx_idct8x8_64_add_c,
                                  &vpx_idct8x8_1_add_neon, TX_8X8, 1),
+                      make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c,
+                                 &vpx_idct4x4_16_add_neon, TX_4X4, 16),
                       make_tuple(&vpx_fdct4x4_c, &vpx_idct4x4_16_add_c,
                                  &vpx_idct4x4_1_add_neon, TX_4X4, 1)));
 #else   // !CONFIG_VP9_HIGHBITDEPTH
diff --git a/vpx_dsp/arm/idct32x32_34_add_neon.c b/vpx_dsp/arm/idct32x32_34_add_neon.c
index ebec9df54..a584b1d9e 100644
--- a/vpx_dsp/arm/idct32x32_34_add_neon.c
+++ b/vpx_dsp/arm/idct32x32_34_add_neon.c
@@ -34,7 +34,7 @@
 // 5 13 20 26
 // 6 21 27 33
 // 7 24 32
-static void idct32_6_neon(const int16_t *input, int16_t *output) {
+static void idct32_6_neon(const tran_low_t *input, int16_t *output) {
   int16x8_t in0, in1, in2, in3, in4, in5, in6, in7;
   int16x8_t s1_0, s1_1, s1_2, s1_3, s1_4, s1_5, s1_6, s1_7, s1_8, s1_9, s1_10,
       s1_11, s1_12, s1_13, s1_14, s1_15, s1_16, s1_17, s1_18, s1_19, s1_20,
@@ -46,8 +46,22 @@ static void idct32_6_neon(const int16_t *input, int16_t *output) {
       s2_31;
   int16x8_t s3_24, s3_25, s3_26, s3_27;
 
-  load_and_transpose_s16_8x8(input, 32, &in0, &in1, &in2, &in3, &in4, &in5,
-                             &in6, &in7);
+  in0 = load_tran_low_to_s16(input);
+  input += 32;
+  in1 = load_tran_low_to_s16(input);
+  input += 32;
+  in2 = load_tran_low_to_s16(input);
+  input += 32;
+  in3 = load_tran_low_to_s16(input);
+  input += 32;
+  in4 = load_tran_low_to_s16(input);
+  input += 32;
+  in5 = load_tran_low_to_s16(input);
+  input += 32;
+  in6 = load_tran_low_to_s16(input);
+  input += 32;
+  in7 = load_tran_low_to_s16(input);
+  transpose_s16_8x8(&in0, &in1, &in2, &in3, &in4, &in5, &in6, &in7);
 
   // stage 1
   // input[1] * cospi_31_64 - input[31] * cospi_1_64 (but input[31] == 0)
@@ -503,7 +517,7 @@ static void idct32_8_neon(const int16_t *input, uint8_t *output, int stride) {
                        output + (24 * stride), stride);
 }
 
-void vpx_idct32x32_34_add_neon(const int16_t *input, uint8_t *dest,
+void vpx_idct32x32_34_add_neon(const tran_low_t *input, uint8_t *dest,
                                int stride) {
   int i;
   int16_t temp[32 * 8];
diff --git a/vpx_dsp/vpx_dsp.mk b/vpx_dsp/vpx_dsp.mk
index 8b292c3a1..bb1143cca 100644
--- a/vpx_dsp/vpx_dsp.mk
+++ b/vpx_dsp/vpx_dsp.mk
@@ -213,7 +213,6 @@ endif  # HAVE_NEON
 endif  # HAVE_NEON_ASM
 DSP_SRCS-$(HAVE_NEON)  += arm/idct16x16_neon.c
 DSP_SRCS-$(HAVE_NEON)  += arm/idct32x32_add_neon.c
-DSP_SRCS-$(HAVE_NEON)  += arm/idct32x32_34_add_neon.c
 DSP_SRCS-$(HAVE_NEON)  += arm/idct32x32_135_add_neon.c
 
 DSP_SRCS-$(HAVE_MSA)   += mips/inv_txfm_msa.h
@@ -246,6 +245,7 @@ DSP_SRCS-$(HAVE_NEON) += arm/idct16x16_1_add_neon.c
 endif  # HAVE_NEON_ASM
 DSP_SRCS-$(HAVE_NEON) += arm/idct_neon.h
 DSP_SRCS-$(HAVE_NEON) += arm/idct32x32_1_add_neon.c
+DSP_SRCS-$(HAVE_NEON) += arm/idct32x32_34_add_neon.c
 
 endif  # CONFIG_VP9
 
diff --git a/vpx_dsp/vpx_dsp_rtcd_defs.pl b/vpx_dsp/vpx_dsp_rtcd_defs.pl
index e884bd321..73da9ef37 100644
--- a/vpx_dsp/vpx_dsp_rtcd_defs.pl
+++ b/vpx_dsp/vpx_dsp_rtcd_defs.pl
@@ -696,7 +696,7 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
     $vpx_idct32x32_135_add_sse2=vpx_idct32x32_1024_add_sse2;
 
     add_proto qw/void vpx_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct32x32_34_add sse2/, "$ssse3_x86_64";
+    specialize qw/vpx_idct32x32_34_add neon sse2/, "$ssse3_x86_64";
 
     add_proto qw/void vpx_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
     specialize qw/vpx_idct32x32_1_add neon sse2/;