avcodec: Implementation of AAC_fixed_decoder (PS-module)

Add fixed point implementation.

Signed-off-by: Nedeljko Babic <nedeljko.babic@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

1 files changed, 165 insertions, 90 deletions

diff --git a/libavcodec/aacps.c b/libavcodec/aacps.c
index ea5a5d2331..bf6047594b 100644
--- a/libavcodec/aacps.c
+++ b/libavcodec/aacps.c
@@ -17,16 +17,23 @@
  * You should have received a copy of the GNU Lesser General Public
  * License along with FFmpeg; if not, write to the Free Software
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ *
+ * Note: Rounding-to-nearest used unless otherwise stated
+ *
  */
 
 #include <stdint.h>
 #include "libavutil/common.h"
-#include "libavutil/internal.h"
 #include "libavutil/mathematics.h"
 #include "avcodec.h"
 #include "get_bits.h"
 #include "aacps.h"
+#if USE_FIXED
+#include "aacps_fixed_tablegen.h"
+#else
+#include "libavutil/internal.h"
 #include "aacps_tablegen.h"
+#endif /* USE_FIXED */
 #include "aacpsdata.c"
 
 #define PS_BASELINE 0  ///< Operate in Baseline PS mode
@@ -148,7 +155,7 @@ static void ipdopd_reset(int8_t *ipd_hist, int8_t *opd_hist)
     }
 }
 
-int ff_ps_read_data(AVCodecContext *avctx, GetBitContext *gb_host, PSContext *ps, int bits_left)
+int AAC_RENAME(ff_ps_read_data)(AVCodecContext *avctx, GetBitContext *gb_host, PSContext *ps, int bits_left)
 {
     int e;
     int bit_count_start = get_bits_count(gb_host);
@@ -302,35 +309,41 @@ err:
 
 /** Split one subband into 2 subsubbands with a symmetric real filter.
  * The filter must have its non-center even coefficients equal to zero. */
-static void hybrid2_re(float (*in)[2], float (*out)[32][2], const float filter[8], int len, int reverse)
+static void hybrid2_re(INTFLOAT (*in)[2], INTFLOAT (*out)[32][2], const INTFLOAT filter[8], int len, int reverse)
 {
     int i, j;
     for (i = 0; i < len; i++, in++) {
-        float re_in = filter[6] * in[6][0];          //real inphase
-        float re_op = 0.0f;                          //real out of phase
-        float im_in = filter[6] * in[6][1];          //imag inphase
-        float im_op = 0.0f;                          //imag out of phase
+        INT64FLOAT re_in = AAC_MUL31(filter[6], in[6][0]); //real inphase
+        INT64FLOAT re_op = 0.0f;                          //real out of phase
+        INT64FLOAT im_in = AAC_MUL31(filter[6], in[6][1]); //imag inphase
+        INT64FLOAT im_op = 0.0f;                          //imag out of phase
         for (j = 0; j < 6; j += 2) {
-            re_op += filter[j+1] * (in[j+1][0] + in[12-j-1][0]);
-            im_op += filter[j+1] * (in[j+1][1] + in[12-j-1][1]);
+            re_op += (INT64FLOAT)filter[j+1] * (in[j+1][0] + in[12-j-1][0]);
+            im_op += (INT64FLOAT)filter[j+1] * (in[j+1][1] + in[12-j-1][1]);
         }
-        out[ reverse][i][0] = re_in + re_op;
-        out[ reverse][i][1] = im_in + im_op;
-        out[!reverse][i][0] = re_in - re_op;
-        out[!reverse][i][1] = im_in - im_op;
+
+#if USE_FIXED
+        re_op = (re_op + 0x40000000) >> 31;
+        im_op = (im_op + 0x40000000) >> 31;
+#endif /* USE_FIXED */
+
+        out[ reverse][i][0] = (INTFLOAT)(re_in + re_op);
+        out[ reverse][i][1] = (INTFLOAT)(im_in + im_op);
+        out[!reverse][i][0] = (INTFLOAT)(re_in - re_op);
+        out[!reverse][i][1] = (INTFLOAT)(im_in - im_op);
     }
 }
 
 /** Split one subband into 6 subsubbands with a complex filter */
-static void hybrid6_cx(PSDSPContext *dsp, float (*in)[2], float (*out)[32][2],
-                       TABLE_CONST float (*filter)[8][2], int len)
+static void hybrid6_cx(PSDSPContext *dsp, INTFLOAT (*in)[2], INTFLOAT (*out)[32][2],
+                       TABLE_CONST INTFLOAT (*filter)[8][2], int len)
 {
     int i;
     int N = 8;
-    LOCAL_ALIGNED_16(float, temp, [8], [2]);
+    LOCAL_ALIGNED_16(INTFLOAT, temp, [8], [2]);
 
     for (i = 0; i < len; i++, in++) {
-        dsp->hybrid_analysis(temp, in, (const float (*)[8][2]) filter, 1, N);
+        dsp->hybrid_analysis(temp, in, (const INTFLOAT (*)[8][2]) filter, 1, N);
         out[0][i][0] = temp[6][0];
         out[0][i][1] = temp[6][1];
         out[1][i][0] = temp[7][0];
@@ -347,18 +360,18 @@ static void hybrid6_cx(PSDSPContext *dsp, float (*in)[2], float (*out)[32][2],
 }
 
 static void hybrid4_8_12_cx(PSDSPContext *dsp,
-                            float (*in)[2], float (*out)[32][2],
-                            TABLE_CONST float (*filter)[8][2], int N, int len)
+                            INTFLOAT (*in)[2], INTFLOAT (*out)[32][2],
+                            TABLE_CONST INTFLOAT (*filter)[8][2], int N, int len)
 {
     int i;
 
     for (i = 0; i < len; i++, in++) {
-        dsp->hybrid_analysis(out[0] + i, in, (const float (*)[8][2]) filter, 32, N);
+        dsp->hybrid_analysis(out[0] + i, in, (const INTFLOAT (*)[8][2]) filter, 32, N);
     }
 }
 
-static void hybrid_analysis(PSDSPContext *dsp, float out[91][32][2],
-                            float in[5][44][2], float L[2][38][64],
+static void hybrid_analysis(PSDSPContext *dsp, INTFLOAT out[91][32][2],
+                            INTFLOAT in[5][44][2], INTFLOAT L[2][38][64],
                             int is34, int len)
 {
     int i, j;
@@ -387,8 +400,8 @@ static void hybrid_analysis(PSDSPContext *dsp, float out[91][32][2],
     }
 }
 
-static void hybrid_synthesis(PSDSPContext *dsp, float out[2][38][64],
-                             float in[91][32][2], int is34, int len)
+static void hybrid_synthesis(PSDSPContext *dsp, INTFLOAT out[2][38][64],
+                             INTFLOAT in[91][32][2], int is34, int len)
 {
     int i, n;
     if (is34) {
@@ -429,7 +442,7 @@ static void hybrid_synthesis(PSDSPContext *dsp, float out[2][38][64],
 }
 
 /// All-pass filter decay slope
-#define DECAY_SLOPE      0.05f
+#define DECAY_SLOPE      Q30(0.05f)
 /// Number of frequency bands that can be addressed by the parameter index, b(k)
 static const int   NR_PAR_BANDS[]      = { 20, 34 };
 static const int   NR_IPDOPD_BANDS[]   = { 11, 17 };
@@ -483,28 +496,43 @@ static void map_idx_34_to_20(int8_t *par_mapped, const int8_t *par, int full)
     }
 }
 
-static void map_val_34_to_20(float par[PS_MAX_NR_IIDICC])
+static void map_val_34_to_20(INTFLOAT par[PS_MAX_NR_IIDICC])
 {
+#if USE_FIXED
+    par[ 0] = (int)(((int64_t)(par[ 0] + (par[ 1]>>1)) * 1431655765 + \
+                      0x40000000) >> 31);
+    par[ 1] = (int)(((int64_t)((par[ 1]>>1) + par[ 2]) * 1431655765 + \
+                      0x40000000) >> 31);
+    par[ 2] = (int)(((int64_t)(par[ 3] + (par[ 4]>>1)) * 1431655765 + \
+                      0x40000000) >> 31);
+    par[ 3] = (int)(((int64_t)((par[ 4]>>1) + par[ 5]) * 1431655765 + \
+                      0x40000000) >> 31);
+#else
     par[ 0] = (2*par[ 0] +   par[ 1]) * 0.33333333f;
     par[ 1] = (  par[ 1] + 2*par[ 2]) * 0.33333333f;
     par[ 2] = (2*par[ 3] +   par[ 4]) * 0.33333333f;
     par[ 3] = (  par[ 4] + 2*par[ 5]) * 0.33333333f;
-    par[ 4] = (  par[ 6] +   par[ 7]) * 0.5f;
-    par[ 5] = (  par[ 8] +   par[ 9]) * 0.5f;
+#endif /* USE_FIXED */
+    par[ 4] = AAC_HALF_SUM(par[ 6], par[ 7]);
+    par[ 5] = AAC_HALF_SUM(par[ 8], par[ 9]);
     par[ 6] =    par[10];
     par[ 7] =    par[11];
-    par[ 8] = (  par[12] +   par[13]) * 0.5f;
-    par[ 9] = (  par[14] +   par[15]) * 0.5f;
+    par[ 8] = AAC_HALF_SUM(par[12], par[13]);
+    par[ 9] = AAC_HALF_SUM(par[14], par[15]);
     par[10] =    par[16];
     par[11] =    par[17];
     par[12] =    par[18];
     par[13] =    par[19];
-    par[14] = (  par[20] +   par[21]) * 0.5f;
-    par[15] = (  par[22] +   par[23]) * 0.5f;
-    par[16] = (  par[24] +   par[25]) * 0.5f;
-    par[17] = (  par[26] +   par[27]) * 0.5f;
+    par[14] = AAC_HALF_SUM(par[20], par[21]);
+    par[15] = AAC_HALF_SUM(par[22], par[23]);
+    par[16] = AAC_HALF_SUM(par[24], par[25]);
+    par[17] = AAC_HALF_SUM(par[26], par[27]);
+#if USE_FIXED
+    par[18] = (((par[28]+2)>>2) + ((par[29]+2)>>2) + ((par[30]+2)>>2) + ((par[31]+2)>>2));
+#else
     par[18] = (  par[28] +   par[29] +   par[30] +   par[31]) * 0.25f;
-    par[19] = (  par[32] +   par[33]) * 0.5f;
+#endif /* USE_FIXED */
+    par[19] = AAC_HALF_SUM(par[32], par[33]);
 }
 
 static void map_idx_10_to_34(int8_t *par_mapped, const int8_t *par, int full)
@@ -589,7 +617,7 @@ static void map_idx_20_to_34(int8_t *par_mapped, const int8_t *par, int full)
     par_mapped[ 0] =  par[ 0];
 }
 
-static void map_val_20_to_34(float par[PS_MAX_NR_IIDICC])
+static void map_val_20_to_34(INTFLOAT par[PS_MAX_NR_IIDICC])
 {
     par[33] =  par[19];
     par[32] =  par[19];
@@ -620,27 +648,29 @@ static void map_val_20_to_34(float par[PS_MAX_NR_IIDICC])
     par[ 7] =  par[ 4];
     par[ 6] =  par[ 4];
     par[ 5] =  par[ 3];
-    par[ 4] = (par[ 2] + par[ 3]) * 0.5f;
+    par[ 4] = AAC_HALF_SUM(par[ 2], par[ 3]);
     par[ 3] =  par[ 2];
     par[ 2] =  par[ 1];
-    par[ 1] = (par[ 0] + par[ 1]) * 0.5f;
+    par[ 1] = AAC_HALF_SUM(par[ 0], par[ 1]);
 }
 
-static void decorrelation(PSContext *ps, float (*out)[32][2], const float (*s)[32][2], int is34)
+static void decorrelation(PSContext *ps, INTFLOAT (*out)[32][2], const INTFLOAT (*s)[32][2], int is34)
 {
-    LOCAL_ALIGNED_16(float, power, [34], [PS_QMF_TIME_SLOTS]);
-    LOCAL_ALIGNED_16(float, transient_gain, [34], [PS_QMF_TIME_SLOTS]);
-    float *peak_decay_nrg = ps->peak_decay_nrg;
-    float *power_smooth = ps->power_smooth;
-    float *peak_decay_diff_smooth = ps->peak_decay_diff_smooth;
-    float (*delay)[PS_QMF_TIME_SLOTS + PS_MAX_DELAY][2] = ps->delay;
-    float (*ap_delay)[PS_AP_LINKS][PS_QMF_TIME_SLOTS + PS_MAX_AP_DELAY][2] = ps->ap_delay;
-    const int8_t *k_to_i = is34 ? k_to_i_34 : k_to_i_20;
-    const float peak_decay_factor = 0.76592833836465f;
+    LOCAL_ALIGNED_16(INTFLOAT, power, [34], [PS_QMF_TIME_SLOTS]);
+    LOCAL_ALIGNED_16(INTFLOAT, transient_gain, [34], [PS_QMF_TIME_SLOTS]);
+    INTFLOAT *peak_decay_nrg = ps->peak_decay_nrg;
+    INTFLOAT *power_smooth = ps->power_smooth;
+    INTFLOAT *peak_decay_diff_smooth = ps->peak_decay_diff_smooth;
+    INTFLOAT (*delay)[PS_QMF_TIME_SLOTS + PS_MAX_DELAY][2] = ps->delay;
+    INTFLOAT (*ap_delay)[PS_AP_LINKS][PS_QMF_TIME_SLOTS + PS_MAX_AP_DELAY][2] = ps->ap_delay;
+#if !USE_FIXED
     const float transient_impact  = 1.5f;
     const float a_smooth          = 0.25f; ///< Smoothing coefficient
+#endif /* USE_FIXED */
+    const int8_t *k_to_i = is34 ? k_to_i_34 : k_to_i_20;
     int i, k, m, n;
     int n0 = 0, nL = 32;
+    const INTFLOAT peak_decay_factor = Q31(0.76592833836465f);;
 
     memset(power, 0, 34 * sizeof(*power));
 
@@ -658,6 +688,33 @@ static void decorrelation(PSContext *ps, float (*out)[32][2], const float (*s)[3
     }
 
     //Transient detection
+#if USE_FIXED
+    for (i = 0; i < NR_PAR_BANDS[is34]; i++) {
+        for (n = n0; n < nL; n++) {
+            int decayed_peak;
+            int denom;
+
+            decayed_peak = (int)(((int64_t)peak_decay_factor * \
+                                           peak_decay_nrg[i] + 0x40000000) >> 31);
+            peak_decay_nrg[i] = FFMAX(decayed_peak, power[i][n]);
+            power_smooth[i] += (power[i][n] - power_smooth[i] + 2) >> 2;
+            peak_decay_diff_smooth[i] += (peak_decay_nrg[i] - power[i][n] - \
+                                          peak_decay_diff_smooth[i] + 2) >> 2;
+            denom = peak_decay_diff_smooth[i] + (peak_decay_diff_smooth[i] >> 1);
+            if (denom > power_smooth[i]) {
+              int p = power_smooth[i];
+              while (denom < 0x40000000) {
+                denom <<= 1;
+                p <<= 1;
+              }
+              transient_gain[i][n] = p / (denom >> 16);
+            }
+            else {
+              transient_gain[i][n] = 1 << 16;
+            }
+        }
+    }
+#else
     for (i = 0; i < NR_PAR_BANDS[is34]; i++) {
         for (n = n0; n < nL; n++) {
             float decayed_peak = peak_decay_factor * peak_decay_nrg[i];
@@ -671,6 +728,7 @@ static void decorrelation(PSContext *ps, float (*out)[32][2], const float (*s)[3
         }
     }
 
+#endif /* USE_FIXED */
     //Decorrelation and transient reduction
     //                         PS_AP_LINKS - 1
     //                               -----
@@ -681,8 +739,22 @@ static void decorrelation(PSContext *ps, float (*out)[32][2], const float (*s)[3
     //d[k][z] (out) = transient_gain_mapped[k][z] * H[k][z] * s[k][z]
     for (k = 0; k < NR_ALLPASS_BANDS[is34]; k++) {
         int b = k_to_i[k];
+#if USE_FIXED
+        int g_decay_slope;
+
+        if (k - DECAY_CUTOFF[is34] <= 0) {
+          g_decay_slope = 1 << 30;
+        }
+        else if (k - DECAY_CUTOFF[is34] >= 20) {
+          g_decay_slope = 0;
+        }
+        else {
+          g_decay_slope = (1 << 30) - DECAY_SLOPE * (k - DECAY_CUTOFF[is34]);
+        }
+#else
         float g_decay_slope = 1.f - DECAY_SLOPE * (k - DECAY_CUTOFF[is34]);
         g_decay_slope = av_clipf(g_decay_slope, 0.f, 1.f);
+#endif /* USE_FIXED */
         memcpy(delay[k], delay[k]+nL, PS_MAX_DELAY*sizeof(delay[k][0]));
         memcpy(delay[k]+PS_MAX_DELAY, s[k], numQMFSlots*sizeof(delay[k][0]));
         for (m = 0; m < PS_AP_LINKS; m++) {
@@ -690,7 +762,7 @@ static void decorrelation(PSContext *ps, float (*out)[32][2], const float (*s)[3
         }
         ps->dsp.decorrelate(out[k], delay[k] + PS_MAX_DELAY - 2, ap_delay[k],
                             phi_fract[is34][k],
-                            (const float (*)[2]) Q_fract_allpass[is34][k],
+                            (const INTFLOAT (*)[2]) Q_fract_allpass[is34][k],
                             transient_gain[b], g_decay_slope, nL - n0);
     }
     for (; k < SHORT_DELAY_BAND[is34]; k++) {
@@ -749,14 +821,14 @@ static void remap20(int8_t (**p_par_mapped)[PS_MAX_NR_IIDICC],
     }
 }
 
-static void stereo_processing(PSContext *ps, float (*l)[32][2], float (*r)[32][2], int is34)
+static void stereo_processing(PSContext *ps, INTFLOAT (*l)[32][2], INTFLOAT (*r)[32][2], int is34)
 {
     int e, b, k;
 
-    float (*H11)[PS_MAX_NUM_ENV+1][PS_MAX_NR_IIDICC] = ps->H11;
-    float (*H12)[PS_MAX_NUM_ENV+1][PS_MAX_NR_IIDICC] = ps->H12;
-    float (*H21)[PS_MAX_NUM_ENV+1][PS_MAX_NR_IIDICC] = ps->H21;
-    float (*H22)[PS_MAX_NUM_ENV+1][PS_MAX_NR_IIDICC] = ps->H22;
+    INTFLOAT (*H11)[PS_MAX_NUM_ENV+1][PS_MAX_NR_IIDICC] = ps->H11;
+    INTFLOAT (*H12)[PS_MAX_NUM_ENV+1][PS_MAX_NR_IIDICC] = ps->H12;
+    INTFLOAT (*H21)[PS_MAX_NUM_ENV+1][PS_MAX_NR_IIDICC] = ps->H21;
+    INTFLOAT (*H22)[PS_MAX_NUM_ENV+1][PS_MAX_NR_IIDICC] = ps->H22;
     int8_t *opd_hist = ps->opd_hist;
     int8_t *ipd_hist = ps->ipd_hist;
     int8_t iid_mapped_buf[PS_MAX_NUM_ENV][PS_MAX_NR_IIDICC];
@@ -768,7 +840,7 @@ static void stereo_processing(PSContext *ps, float (*l)[32][2], float (*r)[32][2
     int8_t (*ipd_mapped)[PS_MAX_NR_IIDICC] = ipd_mapped_buf;
     int8_t (*opd_mapped)[PS_MAX_NR_IIDICC] = opd_mapped_buf;
     const int8_t *k_to_i = is34 ? k_to_i_34 : k_to_i_20;
-    TABLE_CONST float (*H_LUT)[8][4] = (PS_BASELINE || ps->icc_mode < 3) ? HA : HB;
+    TABLE_CONST INTFLOAT (*H_LUT)[8][4] = (PS_BASELINE || ps->icc_mode < 3) ? HA : HB;
 
     //Remapping
     if (ps->num_env_old) {
@@ -823,7 +895,7 @@ static void stereo_processing(PSContext *ps, float (*l)[32][2], float (*r)[32][2
     //Mixing
     for (e = 0; e < ps->num_env; e++) {
         for (b = 0; b < NR_PAR_BANDS[is34]; b++) {
-            float h11, h12, h21, h22;
+            INTFLOAT h11, h12, h21, h22;
             h11 = H_LUT[iid_mapped[e][b] + 7 + 23 * ps->iid_quant][icc_mapped[e][b]][0];
             h12 = H_LUT[iid_mapped[e][b] + 7 + 23 * ps->iid_quant][icc_mapped[e][b]][1];
             h21 = H_LUT[iid_mapped[e][b] + 7 + 23 * ps->iid_quant][icc_mapped[e][b]][2];
@@ -832,27 +904,27 @@ static void stereo_processing(PSContext *ps, float (*l)[32][2], float (*r)[32][2
             if (!PS_BASELINE && ps->enable_ipdopd && b < NR_IPDOPD_BANDS[is34]) {
                 //The spec say says to only run this smoother when enable_ipdopd
                 //is set but the reference decoder appears to run it constantly
-                float h11i, h12i, h21i, h22i;
-                float ipd_adj_re, ipd_adj_im;
+                INTFLOAT h11i, h12i, h21i, h22i;
+                INTFLOAT ipd_adj_re, ipd_adj_im;
                 int opd_idx = opd_hist[b] * 8 + opd_mapped[e][b];
                 int ipd_idx = ipd_hist[b] * 8 + ipd_mapped[e][b];
-                float opd_re = pd_re_smooth[opd_idx];
-                float opd_im = pd_im_smooth[opd_idx];
-                float ipd_re = pd_re_smooth[ipd_idx];
-                float ipd_im = pd_im_smooth[ipd_idx];
+                INTFLOAT opd_re = pd_re_smooth[opd_idx];
+                INTFLOAT opd_im = pd_im_smooth[opd_idx];
+                INTFLOAT ipd_re = pd_re_smooth[ipd_idx];
+                INTFLOAT ipd_im = pd_im_smooth[ipd_idx];
                 opd_hist[b] = opd_idx & 0x3F;
                 ipd_hist[b] = ipd_idx & 0x3F;
 
-                ipd_adj_re = opd_re*ipd_re + opd_im*ipd_im;
-                ipd_adj_im = opd_im*ipd_re - opd_re*ipd_im;
-                h11i = h11 * opd_im;
-                h11  = h11 * opd_re;
-                h12i = h12 * ipd_adj_im;
-                h12  = h12 * ipd_adj_re;
-                h21i = h21 * opd_im;
-                h21  = h21 * opd_re;
-                h22i = h22 * ipd_adj_im;
-                h22  = h22 * ipd_adj_re;
+                ipd_adj_re = AAC_MADD30(opd_re, ipd_re, opd_im, ipd_im);
+                ipd_adj_im = AAC_MSUB30(opd_im, ipd_re, opd_re, ipd_im);
+                h11i = AAC_MUL30(h11,  opd_im);
+                h11  = AAC_MUL30(h11,  opd_re);
+                h12i = AAC_MUL30(h12,  ipd_adj_im);
+                h12  = AAC_MUL30(h12,  ipd_adj_re);
+                h21i = AAC_MUL30(h21,  opd_im);
+                h21  = AAC_MUL30(h21,  opd_re);
+                h22i = AAC_MUL30(h22,  ipd_adj_im);
+                h22  = AAC_MUL30(h22,  ipd_adj_re);
                 H11[1][e+1][b] = h11i;
                 H12[1][e+1][b] = h12i;
                 H21[1][e+1][b] = h21i;
@@ -864,11 +936,14 @@ static void stereo_processing(PSContext *ps, float (*l)[32][2], float (*r)[32][2
             H22[0][e+1][b] = h22;
         }
         for (k = 0; k < NR_BANDS[is34]; k++) {
-            float h[2][4];
-            float h_step[2][4];
+            INTFLOAT h[2][4];
+            INTFLOAT h_step[2][4];
             int start = ps->border_position[e];
             int stop  = ps->border_position[e+1];
-            float width = 1.f / (stop - start);
+            INTFLOAT width = Q30(1.f) / (stop - start);
+#if USE_FIXED
+            width <<= 1;
+#endif
             b = k_to_i[k];
             h[0][0] = H11[0][e][b];
             h[0][1] = H12[0][e][b];
@@ -889,15 +964,15 @@ static void stereo_processing(PSContext *ps, float (*l)[32][2], float (*r)[32][2
             }
             }
             //Interpolation
-            h_step[0][0] = (H11[0][e+1][b] - h[0][0]) * width;
-            h_step[0][1] = (H12[0][e+1][b] - h[0][1]) * width;
-            h_step[0][2] = (H21[0][e+1][b] - h[0][2]) * width;
-            h_step[0][3] = (H22[0][e+1][b] - h[0][3]) * width;
+            h_step[0][0] = AAC_MSUB31_V3(H11[0][e+1][b], h[0][0], width);
+            h_step[0][1] = AAC_MSUB31_V3(H12[0][e+1][b], h[0][1], width);
+            h_step[0][2] = AAC_MSUB31_V3(H21[0][e+1][b], h[0][2], width);
+            h_step[0][3] = AAC_MSUB31_V3(H22[0][e+1][b], h[0][3], width);
             if (!PS_BASELINE && ps->enable_ipdopd) {
-                h_step[1][0] = (H11[1][e+1][b] - h[1][0]) * width;
-                h_step[1][1] = (H12[1][e+1][b] - h[1][1]) * width;
-                h_step[1][2] = (H21[1][e+1][b] - h[1][2]) * width;
-                h_step[1][3] = (H22[1][e+1][b] - h[1][3]) * width;
+                h_step[1][0] = AAC_MSUB31_V3(H11[1][e+1][b], h[1][0], width);
+                h_step[1][1] = AAC_MSUB31_V3(H12[1][e+1][b], h[1][1], width);
+                h_step[1][2] = AAC_MSUB31_V3(H21[1][e+1][b], h[1][2], width);
+                h_step[1][3] = AAC_MSUB31_V3(H22[1][e+1][b], h[1][3], width);
             }
             ps->dsp.stereo_interpolate[!PS_BASELINE && ps->enable_ipdopd](
                 l[k] + start + 1, r[k] + start + 1,
@@ -906,10 +981,10 @@ static void stereo_processing(PSContext *ps, float (*l)[32][2], float (*r)[32][2
     }
 }
 
-int ff_ps_apply(AVCodecContext *avctx, PSContext *ps, float L[2][38][64], float R[2][38][64], int top)
+int AAC_RENAME(ff_ps_apply)(AVCodecContext *avctx, PSContext *ps, INTFLOAT L[2][38][64], INTFLOAT R[2][38][64], int top)
 {
-    float (*Lbuf)[32][2] = ps->Lbuf;
-    float (*Rbuf)[32][2] = ps->Rbuf;
+    INTFLOAT (*Lbuf)[32][2] = ps->Lbuf;
+    INTFLOAT (*Rbuf)[32][2] = ps->Rbuf;
     const int len = 32;
     int is34 = ps->is34bands;
 
@@ -919,7 +994,7 @@ int ff_ps_apply(AVCodecContext *avctx, PSContext *ps, float L[2][38][64], float
         memset(ps->ap_delay + top, 0, (NR_ALLPASS_BANDS[is34] - top)*sizeof(ps->ap_delay[0]));
 
     hybrid_analysis(&ps->dsp, Lbuf, ps->in_buf, L, is34, len);
-    decorrelation(ps, Rbuf, (const float (*)[32][2]) Lbuf, is34);
+    decorrelation(ps, Rbuf, (const INTFLOAT (*)[32][2]) Lbuf, is34);
     stereo_processing(ps, Lbuf, Rbuf, is34);
     hybrid_synthesis(&ps->dsp, L, Lbuf, is34, len);
     hybrid_synthesis(&ps->dsp, R, Rbuf, is34, len);
@@ -936,7 +1011,7 @@ int ff_ps_apply(AVCodecContext *avctx, PSContext *ps, float L[2][38][64], float
 #define PS_VLC_ROW(name) \
     { name ## _codes, name ## _bits, sizeof(name ## _codes), sizeof(name ## _codes[0]) }
 
-av_cold void ff_ps_init(void) {
+av_cold void AAC_RENAME(ff_ps_init)(void) {
     // Syntax initialization
     static const struct {
         const void *ps_codes, *ps_bits;
@@ -968,7 +1043,7 @@ av_cold void ff_ps_init(void) {
     ps_tableinit();
 }
 
-av_cold void ff_ps_ctx_init(PSContext *ps)
+av_cold void AAC_RENAME(ff_ps_ctx_init)(PSContext *ps)
 {
-    ff_psdsp_init(&ps->dsp);
+    AAC_RENAME(ff_psdsp_init)(&ps->dsp);
 }