diff --git a/src/audio/mixin_mixout/CMakeLists.txt b/src/audio/mixin_mixout/CMakeLists.txt
index b1723cd114ae..2e6dc7456ae5 100644
--- a/src/audio/mixin_mixout/CMakeLists.txt
+++ b/src/audio/mixin_mixout/CMakeLists.txt
@@ -1 +1 @@
-add_local_sources(sof mixin_mixout.c mixin_mixout_generic.c mixin_mixout_hifi3.c)
+add_local_sources(sof mixin_mixout.c mixin_mixout_generic.c mixin_mixout_hifi3.c mixin_mixout_hifi5.c)
diff --git a/src/audio/mixin_mixout/Kconfig b/src/audio/mixin_mixout/Kconfig
index 449d6617fce7..2315023d61d1 100644
--- a/src/audio/mixin_mixout/Kconfig
+++ b/src/audio/mixin_mixout/Kconfig
@@ -6,3 +6,40 @@ config COMP_MIXIN_MIXOUT
 	default y
 	help
 	  Select for Mixin_mixout component
+
+choice "MIXIN_MIXOUT_SIMD_LEVEL_SELECT"
+	prompt "choose which SIMD level used for MIXIN_MIXOUT module"
+	depends on COMP_MIXIN_MIXOUT
+	default MIXIN_MIXOUT_HIFI_MAX
+
+	config MIXIN_MIXOUT_HIFI_MAX
+		prompt "Max HiFi level available in the toolchain"
+		bool
+		help
+			When this was selected, optimization level will be determined
+			by toolchain.
+
+	config MIXIN_MIXOUT_HIFI_5
+		prompt "choose HIFI4 intrinsic optimized MIXIN_MIXOUT module"
+		bool
+		help
+			This option used to build HIFI4 optimized MIXIN_MIXOUT code
+
+	config MIXIN_MIXOUT_HIFI_4
+		prompt "choose HIFI4 intrinsic optimized MIXIN_MIXOUT module"
+		bool
+		help
+			This option used to build HIFI4 optimized MIXIN_MIXOUT code
+
+	config MIXIN_MIXOUT_HIFI_3
+		prompt "choose HIFI3 intrinsic optimized MIXIN_MIXOUT module"
+		bool
+		help
+			This option used to build HIFI3 intrinsic optimized MIXIN_MIXOUT code
+
+	config MIXIN_MIXOUT_HIFI_NONE
+		prompt "choose generic C MIXIN_MIXOUT module, no HIFI SIMD involved"
+		bool
+		help
+			This option used to build MIXIN_MIXOUT generic code.
+endchoice
diff --git a/src/audio/mixin_mixout/mixin_mixout.h b/src/audio/mixin_mixout/mixin_mixout.h
index 8ac844383917..085187e0e743 100644
--- a/src/audio/mixin_mixout/mixin_mixout.h
+++ b/src/audio/mixin_mixout/mixin_mixout.h
@@ -31,18 +31,6 @@
 #include <sof/platform.h>
 #include <stddef.h>
 
-#define MIXIN_MIXOUT_GENERIC
-
-#if defined(__XCC__)
-
-#include <xtensa/config/core-isa.h>
-#if XCHAL_HAVE_HIFI3 || XCHAL_HAVE_HIFI4
-#undef MIXIN_MIXOUT_GENERIC
-#define MIXIN_MIXOUT_HIFI3
-#endif
-
-#endif
-
 enum ipc4_mixin_config_param {
 	/* large_config_set param id for ipc4_mixer_mode_config */
 	IPC4_MIXER_MODE = 1
diff --git a/src/audio/mixin_mixout/mixin_mixout_generic.c b/src/audio/mixin_mixout/mixin_mixout_generic.c
index c69305073660..60a9f8211b05 100644
--- a/src/audio/mixin_mixout/mixin_mixout_generic.c
+++ b/src/audio/mixin_mixout/mixin_mixout_generic.c
@@ -9,7 +9,7 @@
 
 #include "mixin_mixout.h"
 
-#ifdef MIXIN_MIXOUT_GENERIC
+#if SOF_USE_HIFI(NONE, MIXIN_MIXOUT)
 
 #if CONFIG_FORMAT_S16LE
 static void mix_s16(struct cir_buf_ptr *sink, int32_t start_sample, int32_t mixed_samples,
diff --git a/src/audio/mixin_mixout/mixin_mixout_hifi3.c b/src/audio/mixin_mixout/mixin_mixout_hifi3.c
index 369ede78d36a..568cbed5202b 100644
--- a/src/audio/mixin_mixout/mixin_mixout_hifi3.c
+++ b/src/audio/mixin_mixout/mixin_mixout_hifi3.c
@@ -8,7 +8,7 @@
 
 #include "mixin_mixout.h"
 
-#ifdef MIXIN_MIXOUT_HIFI3
+#if SOF_USE_HIFI(3, MIXIN_MIXOUT) || SOF_USE_HIFI(4, MIXIN_MIXOUT)
 
 #if CONFIG_FORMAT_S16LE
 static void mix_s16(struct cir_buf_ptr *sink, int32_t start_sample, int32_t mixed_samples,
diff --git a/src/audio/mixin_mixout/mixin_mixout_hifi5.c b/src/audio/mixin_mixout/mixin_mixout_hifi5.c
new file mode 100644
index 000000000000..6d82ec54d50d
--- /dev/null
+++ b/src/audio/mixin_mixout/mixin_mixout_hifi5.c
@@ -0,0 +1,285 @@
+// SPDX-License-Identifier: BSD-3-Clause
+//
+// Copyright(c) 2024 Intel Corporation. All rights reserved.
+//
+// Author: Andrula Song <xiaoyuan.song@intel.com>
+
+#include <sof/common.h>
+
+#include "mixin_mixout.h"
+
+#if SOF_USE_HIFI(5, MIXIN_MIXOUT)
+
+#if CONFIG_FORMAT_S16LE
+static void mix_s16(struct cir_buf_ptr *sink, int32_t start_sample, int32_t mixed_samples,
+		    const struct cir_buf_ptr *source,
+		    int32_t sample_count, uint16_t gain)
+{
+	int samples_to_mix, samples_to_copy, left_samples;
+	int n, nmax, i, m, left;
+	ae_int16x4 in_sample, in_sample1;
+	ae_int16x4 out_sample, out_sample1;
+	ae_int16x8 *in;
+	ae_int16x8 *out;
+	ae_valignx2 inu = AE_ZALIGN128();
+	ae_valignx2 outu1 = AE_ZALIGN128();
+	ae_valignx2 outu2 = AE_ZALIGN128();
+	/* cir_buf_wrap() is required and is done below in a loop */
+	ae_int16 *dst = (ae_int16 *)sink->ptr + start_sample;
+	ae_int16 *src = source->ptr;
+
+	assert(mixed_samples >= start_sample);
+	samples_to_mix = AE_MIN32(mixed_samples - start_sample, sample_count);
+	samples_to_copy = sample_count - samples_to_mix;
+	n = 0;
+
+	for (left_samples = samples_to_mix; left_samples > 0; left_samples -= n) {
+		src = cir_buf_wrap(src + n, source->buf_start, source->buf_end);
+		dst = cir_buf_wrap(dst + n, sink->buf_start, sink->buf_end);
+		/* calculate the remaining samples*/
+		nmax = (ae_int16 *)source->buf_end - src;
+		n = AE_MIN32(left_samples, nmax);
+		nmax = (ae_int16 *)sink->buf_end - dst;
+		n = AE_MIN32(n, nmax);
+		in = (ae_int16x8 *)src;
+		out = (ae_int16x8 *)dst;
+		inu = AE_LA128_PP(in);
+		outu1 = AE_LA128_PP(out);
+		m = n >> 3;
+		left = n & 0x07;
+		/* process 8 samples per loop */
+		for (i = 0; i < m; i++) {
+			AE_LA16X4X2_IP(in_sample, in_sample1, inu, in);
+			AE_LA16X4X2_IP(out_sample, out_sample1, outu1, out);
+			out--;
+			out_sample = AE_ADD16S(in_sample, out_sample);
+			out_sample1 = AE_ADD16S(in_sample1, out_sample1);
+			AE_SA16X4X2_IP(out_sample, out_sample1, outu2, out);
+		}
+		AE_SA128POS_FP(outu2, out);
+
+		/* process the left samples that less than 8
+		 * one by one to avoid memory access overrun
+		 */
+		for (i = 0; i < left ; i++) {
+			AE_L16_IP(in_sample, (ae_int16 *)in, sizeof(ae_int16));
+			AE_L16_IP(out_sample, (ae_int16 *)out, 0);
+			out_sample = AE_ADD16S(in_sample, out_sample);
+			AE_S16_0_IP(out_sample, (ae_int16 *)out, sizeof(ae_int16));
+		}
+	}
+
+	for (left_samples = samples_to_copy; left_samples > 0; left_samples -= n) {
+		src = cir_buf_wrap(src + n, source->buf_start, source->buf_end);
+		dst = cir_buf_wrap(dst + n, sink->buf_start, sink->buf_end);
+		/* calculate the remaining samples*/
+		nmax = (ae_int16 *)source->buf_end - src;
+		n = AE_MIN32(left_samples, nmax);
+		nmax = (ae_int16 *)sink->buf_end - dst;
+		n = AE_MIN32(n, nmax);
+		in = (ae_int16x8 *)src;
+		out = (ae_int16x8 *)dst;
+		inu = AE_LA128_PP(in);
+		m = n >> 3;
+		left = n & 0x07;
+		/* process 8 frames per loop */
+		for (i = 0; i < m; i++) {
+			AE_LA16X4X2_IP(in_sample, in_sample1, inu, in);
+			AE_SA16X4X2_IP(in_sample, in_sample1, outu2, out);
+		}
+		AE_SA128POS_FP(outu2, out);
+
+		/* process the left samples that less than 8
+		 * one by one to avoid memory access overrun
+		 */
+		for (i = 0; i < left ; i++) {
+			AE_L16_IP(in_sample, (ae_int16 *)in, sizeof(ae_int16));
+			AE_S16_0_IP(in_sample, (ae_int16 *)out, sizeof(ae_int16));
+		}
+	}
+}
+#endif	/* CONFIG_FORMAT_S16LE */
+
+#if CONFIG_FORMAT_S24LE
+static void mix_s24(struct cir_buf_ptr *sink, int32_t start_sample, int32_t mixed_samples,
+		    const struct cir_buf_ptr *source,
+		    int32_t sample_count, uint16_t gain)
+{
+	int samples_to_mix, samples_to_copy, left_samples;
+	int n, nmax, i, m, left;
+	ae_int32x2 in_sample, in_sample1;
+	ae_int32x2 out_sample, out_sample1;
+	ae_int32x4 *in;
+	ae_int32x4 *out;
+	ae_valignx2 inu = AE_ZALIGN128();
+	ae_valignx2 outu1 = AE_ZALIGN128();
+	ae_valignx2 outu2 = AE_ZALIGN128();
+	/* cir_buf_wrap() is required and is done below in a loop */
+	int32_t *dst = (int32_t *)sink->ptr + start_sample;
+	int32_t *src = source->ptr;
+
+	assert(mixed_samples >= start_sample);
+	samples_to_mix = AE_MIN32(mixed_samples - start_sample, sample_count);
+	samples_to_copy = sample_count - samples_to_mix;
+	n = 0;
+
+	for (left_samples = samples_to_mix; left_samples > 0; left_samples -= n) {
+		src = cir_buf_wrap(src + n, source->buf_start, source->buf_end);
+		dst = cir_buf_wrap(dst + n, sink->buf_start, sink->buf_end);
+		/* calculate the remaining samples*/
+		nmax = (int32_t *)source->buf_end - src;
+		n = AE_MIN32(left_samples, nmax);
+		nmax = (int32_t *)sink->buf_end - dst;
+		n = AE_MIN32(n, nmax);
+		in = (ae_int32x4 *)src;
+		out = (ae_int32x4 *)dst;
+		inu = AE_LA128_PP(in);
+		outu1 = AE_LA128_PP(out);
+		m = n >> 2;
+		left = n & 3;
+		/* process 2 samples per time */
+		for (i = 0; i < m; i++) {
+			AE_LA32X2X2_IP(in_sample, in_sample1, inu, in);
+			AE_LA32X2X2_IP(out_sample, out_sample1, outu1, out);
+			out--;
+			out_sample = AE_ADD24S(in_sample, out_sample);
+			out_sample1 = AE_ADD24S(in_sample1, out_sample1);
+			AE_SA32X2X2_IP(out_sample, out_sample1, outu2, out);
+		}
+		AE_SA128POS_FP(outu2, out);
+
+		/* process the left sample to avoid memory access overrun */
+		if (left) {
+			AE_L32_IP(in_sample, (ae_int32 *)in, sizeof(ae_int32));
+			AE_L32_IP(out_sample, (ae_int32 *)out, 0);
+			out_sample = AE_ADD24S(in_sample, out_sample);
+			AE_S32_L_IP(out_sample, (ae_int32 *)out, sizeof(ae_int32));
+		}
+	}
+
+	for (left_samples = samples_to_copy; left_samples > 0; left_samples -= n) {
+		src = cir_buf_wrap(src + n, source->buf_start, source->buf_end);
+		dst = cir_buf_wrap(dst + n, sink->buf_start, sink->buf_end);
+		nmax = (int32_t *)source->buf_end - src;
+		n = AE_MIN32(left_samples, nmax);
+		nmax = (int32_t *)sink->buf_end - dst;
+		n = AE_MIN32(n, nmax);
+		in = (ae_int32x4 *)src;
+		out = (ae_int32x4 *)dst;
+		inu = AE_LA128_PP(in);
+		m = n >> 2;
+		left = n & 3;
+		for (i = 0; i < m; i++) {
+			AE_LA32X2X2_IP(in_sample, in_sample1, inu, in);
+			AE_SA32X2X2_IP(in_sample, in_sample1, outu2, out);
+		}
+		AE_SA128POS_FP(outu2, out);
+		/* process the left sample to avoid memory access overrun */
+		if (left) {
+			AE_L32_IP(in_sample, (ae_int32 *)in, sizeof(ae_int32));
+			AE_S32_L_IP(in_sample, (ae_int32 *)out, sizeof(ae_int32));
+		}
+	}
+}
+
+#endif	/* CONFIG_FORMAT_S24LE */
+
+#if CONFIG_FORMAT_S32LE
+static void mix_s32(struct cir_buf_ptr *sink, int32_t start_sample, int32_t mixed_samples,
+		    const struct cir_buf_ptr *source,
+		    int32_t sample_count, uint16_t gain)
+{
+	int samples_to_mix, samples_to_copy, left_samples;
+	int n, nmax, i, m, left;
+	ae_int32x2 in_sample, in_sample1;
+	ae_int32x2 out_sample, out_sample1;
+	ae_int32x4 *in;
+	ae_int32x4 *out;
+	ae_valignx2 inu = AE_ZALIGN128();
+	ae_valignx2 outu1 = AE_ZALIGN128();
+	ae_valignx2 outu2 = AE_ZALIGN128();
+	/* cir_buf_wrap() is required and is done below in a loop */
+	int32_t *dst = (int32_t *)sink->ptr + start_sample;
+	int32_t *src = source->ptr;
+
+	assert(mixed_samples >= start_sample);
+	samples_to_mix = AE_MIN32(mixed_samples - start_sample, sample_count);
+	samples_to_copy = sample_count - samples_to_mix;
+	n = 0;
+
+	for (left_samples = samples_to_mix; left_samples > 0; left_samples -= n) {
+		src = cir_buf_wrap(src + n, source->buf_start, source->buf_end);
+		dst = cir_buf_wrap(dst + n, sink->buf_start, sink->buf_end);
+		/* calculate the remaining samples*/
+		nmax = (int32_t *)source->buf_end - src;
+		n = AE_MIN32(left_samples, nmax);
+		nmax = (int32_t *)sink->buf_end - dst;
+		n = AE_MIN32(n, nmax);
+		in = (ae_int32x4 *)src;
+		out = (ae_int32x4 *)dst;
+		inu = AE_LA128_PP(in);
+		outu1 = AE_LA128_PP(out);
+		m = n >> 2;
+		left = n & 3;
+		for (i = 0; i < m; i++) {
+			AE_LA32X2X2_IP(in_sample, in_sample1, inu, in);
+			AE_LA32X2X2_IP(out_sample, out_sample1, outu1, out);
+			out--;
+			out_sample = AE_ADD32S(in_sample, out_sample);
+			out_sample1 = AE_ADD32S(in_sample1, out_sample1);
+			AE_SA32X2X2_IP(out_sample, out_sample1, outu2, out);
+		}
+		AE_SA128POS_FP(outu2, out);
+
+		/* process the left sample to avoid memory access overrun */
+		if (left) {
+			AE_L32_IP(in_sample, (ae_int32 *)in, sizeof(ae_int32));
+			AE_L32_IP(out_sample, (ae_int32 *)out, 0);
+			out_sample = AE_ADD32S(in_sample, out_sample);
+			AE_S32_L_IP(out_sample, (ae_int32 *)out, sizeof(ae_int32));
+		}
+	}
+
+	for (left_samples = samples_to_copy; left_samples > 0; left_samples -= n) {
+		src = cir_buf_wrap(src + n, source->buf_start, source->buf_end);
+		dst = cir_buf_wrap(dst + n, sink->buf_start, sink->buf_end);
+		/* calculate the remaining samples*/
+		nmax = (int32_t *)source->buf_end - src;
+		n = AE_MIN32(left_samples, nmax);
+		nmax = (int32_t *)sink->buf_end - dst;
+		n = AE_MIN32(n, nmax);
+		in = (ae_int32x4 *)src;
+		out = (ae_int32x4 *)dst;
+		inu = AE_LA128_PP(in);
+		m = n >> 2;
+		left = n & 3;
+		for (i = 0; i < m; i++) {
+			AE_LA32X2X2_IP(in_sample, in_sample1, inu, in);
+			AE_SA32X2X2_IP(in_sample, in_sample1, outu2, out);
+		}
+		AE_SA128POS_FP(outu2, out);
+		/* process the left sample to avoid memory access overrun */
+		if (left) {
+			AE_L32_IP(in_sample, (ae_int32 *)in, sizeof(ae_int32));
+			AE_S32_L_IP(in_sample, (ae_int32 *)out, sizeof(ae_int32));
+		}
+	}
+}
+
+#endif	/* CONFIG_FORMAT_S32LE */
+
+const struct mix_func_map mix_func_map[] = {
+#if CONFIG_FORMAT_S16LE
+	{ SOF_IPC_FRAME_S16_LE, mix_s16 },
+#endif
+#if CONFIG_FORMAT_S24LE
+	{ SOF_IPC_FRAME_S24_4LE, mix_s24 },
+#endif
+#if CONFIG_FORMAT_S32LE
+	{ SOF_IPC_FRAME_S32_LE, mix_s32 }
+#endif
+};
+
+const size_t mix_count = ARRAY_SIZE(mix_func_map);
+
+#endif
diff --git a/zephyr/CMakeLists.txt b/zephyr/CMakeLists.txt
index c2cf0bcfffb1..5a871a1ab027 100644
--- a/zephyr/CMakeLists.txt
+++ b/zephyr/CMakeLists.txt
@@ -561,6 +561,7 @@ zephyr_library_sources_ifdef(CONFIG_COMP_MIXIN_MIXOUT
 	${SOF_AUDIO_PATH}/mixin_mixout/mixin_mixout.c
 	${SOF_AUDIO_PATH}/mixin_mixout/mixin_mixout_generic.c
 	${SOF_AUDIO_PATH}/mixin_mixout/mixin_mixout_hifi3.c
+	${SOF_AUDIO_PATH}/mixin_mixout/mixin_mixout_hifi5.c
 )
 
 zephyr_library_sources_ifdef(CONFIG_COMP_TONE