Upgrade to FDKv2

Bug: 71430241
Test: CTS DecoderTest and DecoderTestAacDrc

original-Change-Id: Iaa20f749b8a04d553b20247cfe1a8930ebbabe30

Apply clang-format also on header files.

original-Change-Id: I14de1ef16bbc79ec0283e745f98356a10efeb2e4

Fixes for MPEG-D DRC

original-Change-Id: If1de2d74bbbac84b3f67de3b88b83f6a23b8a15c

Catch unsupported tw_mdct at an early stage

original-Change-Id: Ied9dd00d754162a0e3ca1ae3e6b854315d818afe

Fixing PVC transition frames

original-Change-Id: Ib75725abe39252806c32d71176308f2c03547a4e

Move qmf bands sanity check

original-Change-Id: Iab540c3013c174d9490d2ae100a4576f51d8dbc4

Initialize scaling variable

original-Change-Id: I3c4087101b70e998c71c1689b122b0d7762e0f9e

Add 16 qmf band configuration to getSlotNrgHQ()

original-Change-Id: I49a5d30f703a1b126ff163df9656db2540df21f1

Always apply byte alignment at the end of the AudioMuxElement

original-Change-Id: I42d560287506d65d4c3de8bfe3eb9a4ebeb4efc7

Setup SBR element only if no parse error exists

original-Change-Id: I1915b73704bc80ab882b9173d6bec59cbd073676

Additional array index check in HCR

original-Change-Id: I18cc6e501ea683b5009f1bbee26de8ddd04d8267

Fix fade-in index selection in concealment module

original-Change-Id: Ibf802ed6ed8c05e9257e1f3b6d0ac1162e9b81c1

Enable explicit backward compatible parser for AAC_LD

original-Change-Id: I27e9c678dcb5d40ed760a6d1e06609563d02482d

Skip spatial specific config in explicit backward compatible ASC

original-Change-Id: Iff7cc365561319e886090cedf30533f562ea4d6e

Update flags description in decoder API

original-Change-Id: I9a5b4f8da76bb652f5580cbd3ba9760425c43830

Add QMF domain reset function

original-Change-Id: I4f89a8a2c0277d18103380134e4ed86996e9d8d6

DRC upgrade v2.1.0

original-Change-Id: I5731c0540139dab220094cd978ef42099fc45b74

Fix integer overflow in sqrtFixp_lookup()

original-Change-Id: I429a6f0d19aa2cc957e0f181066f0ca73968c914

Fix integer overflow in invSqrtNorm2()

original-Change-Id: I84de5cbf9fb3adeb611db203fe492fabf4eb6155

Fix integer overflow in GenerateRandomVector()

original-Change-Id: I3118a641008bd9484d479e5b0b1ee2b5d7d44d74

Fix integer overflow in adjustTimeSlot_EldGrid()

original-Change-Id: I29d503c247c5c8282349b79df940416a512fb9d5

Fix integer overflow in FDKsbrEnc_codeEnvelope()

original-Change-Id: I6b34b61ebb9d525b0c651ed08de2befc1f801449

Follow-up on: Fix integer overflow in adjustTimeSlot_EldGrid()

original-Change-Id: I6f8f578cc7089e5eb7c7b93e580b72ca35ad689a

Fix integer overflow in get_pk_v2()

original-Change-Id: I63375bed40d45867f6eeaa72b20b1f33e815938c

Fix integer overflow in Syn_filt_zero()

original-Change-Id: Ie0c02fdfbe03988f9d3b20d10cd9fe4c002d1279

Fix integer overflow in CFac_CalcFacSignal()

original-Change-Id: Id2d767c40066c591b51768e978eb8af3b803f0c5

Fix integer overflow in FDKaacEnc_FDKaacEnc_calcPeNoAH()

original-Change-Id: Idcbd0f4a51ae2550ed106aa6f3d678d1f9724841

Fix integer overflow in sbrDecoder_calculateGainVec()

original-Change-Id: I7081bcbe29c5cede9821b38d93de07c7add2d507

Fix integer overflow in CLpc_SynthesisLattice()

original-Change-Id: I4a95ddc18de150102352d4a1845f06094764c881

Fix integer overflow in Pred_Lt4()

original-Change-Id: I4dbd012b2de7d07c3e70a47b92e3bfae8dbc750a

Fix integer overflow in FDKsbrEnc_InitSbrFastTransientDetector()

original-Change-Id: I788cbec1a4a00f44c2f3a72ad7a4afa219807d04

Fix unsigned integer overflow in FDKaacEnc_WriteBitstream()

original-Change-Id: I68fc75166e7d2cd5cd45b18dbe3d8c2a92f1822a

Fix unsigned integer overflow in FDK_MetadataEnc_Init()

original-Change-Id: Ie8d025f9bcdb2442c704bd196e61065c03c10af4

Fix overflow in pseudo random number generators

original-Change-Id: I3e2551ee01356297ca14e3788436ede80bd5513c

Fix unsigned integer overflow in sbrDecoder_Parse()

original-Change-Id: I3f231b2f437e9c37db4d5b964164686710eee971

Fix unsigned integer overflow in longsub()

original-Change-Id: I73c2bc50415cac26f1f5a29e125bbe75f9180a6e

Fix unsigned integer overflow in CAacDecoder_DecodeFrame()

original-Change-Id: Ifce2db4b1454b46fa5f887e9d383f1cc43b291e4

Fix overflow at CLpdChannelStream_Read()

original-Change-Id: Idb9d822ce3a4272e4794b643644f5434e2d4bf3f

Fix unsigned integer overflow in Hcr_State_BODY_SIGN_ESC__ESC_WORD()

original-Change-Id: I1ccf77c0015684b85534c5eb97162740a870b71c

Fix unsigned integer overflow in UsacConfig_Parse()

original-Change-Id: Ie6d27f84b6ae7eef092ecbff4447941c77864d9f

Fix unsigned integer overflow in aacDecoder_drcParse()

original-Change-Id: I713f28e883eea3d70b6fa56a7b8f8c22bcf66ca0

Fix unsigned integer overflow in aacDecoder_drcReadCompression()

original-Change-Id: Ia34dfeb88c4705c558bce34314f584965cafcf7a

Fix unsigned integer overflow in CDataStreamElement_Read()

original-Change-Id: Iae896cc1d11f0a893d21be6aa90bd3e60a2c25f0

Fix unsigned integer overflow in transportDec_AdjustEndOfAccessUnit()

original-Change-Id: I64cf29a153ee784bb4a16fdc088baabebc0007dc

Fix unsigned integer overflow in transportDec_GetAuBitsRemaining()

original-Change-Id: I975b3420faa9c16a041874ba0db82e92035962e4

Fix unsigned integer overflow in extractExtendedData()

original-Change-Id: I2a59eb09e2053cfb58dfb75fcecfad6b85a80a8f

Fix signed integer overflow in CAacDecoder_ExtPayloadParse()

original-Change-Id: I4ad5ca4e3b83b5d964f1c2f8c5e7b17c477c7929

Fix unsigned integer overflow in CAacDecoder_DecodeFrame()

original-Change-Id: I29a39df77d45c52a0c9c5c83c1ba81f8d0f25090

Follow-up on: Fix integer overflow in CLpc_SynthesisLattice()

original-Change-Id: I8fb194ffc073a3432a380845be71036a272d388f

Fix signed integer overflow in _interpolateDrcGain()

original-Change-Id: I879ec9ab14005069a7c47faf80e8bc6e03d22e60

Fix unsigned integer overflow in FDKreadBits()

original-Change-Id: I1f47a6a8037ff70375aa8844947d5681bb4287ad

Fix unsigned integer overflow in FDKbyteAlign()

original-Change-Id: Id5f3a11a0c9e50fc6f76ed6c572dbd4e9f2af766

Fix unsigned integer overflow in FDK_get32()

original-Change-Id: I9d33b8e97e3d38cbb80629cb859266ca0acdce96

Fix unsigned integer overflow in FDK_pushBack()

original-Change-Id: Ic87f899bc8c6acf7a377a8ca7f3ba74c3a1e1c19

Fix unsigned integer overflow in FDK_pushForward()

original-Change-Id: I3b754382f6776a34be1602e66694ede8e0b8effc

Fix unsigned integer overflow in ReadPsData()

original-Change-Id: I25361664ba8139e32bbbef2ca8c106a606ce9c37

Fix signed integer overflow in E_UTIL_residu()

original-Change-Id: I8c3abd1f437ee869caa8fb5903ce7d3d641b6aad

REVERT: Follow-up on: Integer overflow in CLpc_SynthesisLattice().

original-Change-Id: I3d340099acb0414795c8dfbe6362bc0a8f045f9b

Follow-up on: Fix integer overflow in CLpc_SynthesisLattice()

original-Change-Id: I4aedb8b3a187064e9f4d985175aa55bb99cc7590

Follow-up on: Fix unsigned integer overflow in aacDecoder_drcParse()

original-Change-Id: I2aa2e13916213bf52a67e8b0518e7bf7e57fb37d

Fix integer overflow in acelp

original-Change-Id: Ie6390c136d84055f8b728aefbe4ebef6e029dc77

Fix unsigned integer overflow in aacDecoder_UpdateBitStreamCounters()

original-Change-Id: I391ffd97ddb0b2c184cba76139bfb356a3b4d2e2

Adjust concealment default settings

original-Change-Id: I6a95db935a327c47df348030bcceafcb29f54b21

Saturate estimatedStartPos

original-Change-Id: I27be2085e0ae83ec9501409f65e003f6bcba1ab6

Negative shift exponent in _interpolateDrcGain()

original-Change-Id: I18edb26b26d002aafd5e633d4914960f7a359c29

Negative shift exponent in calculateICC()

original-Change-Id: I3dcd2ae98d2eb70ee0d59750863cbb2a6f4f8aba

Too large shift exponent in FDK_put()

original-Change-Id: Ib7d9aaa434d2d8de4a13b720ca0464b31ca9b671

Too large shift exponent in CalcInvLdData()

original-Change-Id: I43e6e78d4cd12daeb1dcd5d82d1798bdc2550262

Member access within null pointer of type SBR_CHANNEL

original-Change-Id: Idc5e4ea8997810376d2f36bbdf628923b135b097

Member access within null pointer of type CpePersistentData

original-Change-Id: Ib6c91cb0d37882768e5baf63324e429589de0d9d

Member access within null pointer FDKaacEnc_psyMain()

original-Change-Id: I7729b7f4479970531d9dc823abff63ca52e01997

Member access within null pointer FDKaacEnc_GetPnsParam()

original-Change-Id: I9aa3b9f3456ae2e0f7483dbd5b3dde95fc62da39

Member access within null pointer FDKsbrEnc_EnvEncodeFrame()

original-Change-Id: I67936f90ea714e90b3e81bc0dd1472cc713eb23a

Add HCR sanity check

original-Change-Id: I6c1d9732ebcf6af12f50b7641400752f74be39f7

Fix memory issue for HBE edge case with 8:3 SBR

original-Change-Id: I11ea58a61e69fbe8bf75034b640baee3011e63e9

Additional SBR parametrization sanity check for ELD

original-Change-Id: Ie26026fbfe174c2c7b3691f6218b5ce63e322140

Add MPEG-D DRC channel layout check

original-Change-Id: Iea70a74f171b227cce636a9eac4ba662777a2f72

Additional out-of-bounds checks in MPEG-D DRC

original-Change-Id: Ife4a8c3452c6fde8a0a09e941154a39a769777d4

Change-Id: Ic63cb2f628720f54fe9b572b0cb528e2599c624e

1 files changed, 405 insertions, 0 deletions

diff --git a/libFDK/include/qmf_pcm.h b/libFDK/include/qmf_pcm.h
new file mode 100644
index 0000000..f24e0cd
--- /dev/null
+++ b/libFDK/include/qmf_pcm.h
@@ -0,0 +1,405 @@
+/* -----------------------------------------------------------------------------
+Software License for The Fraunhofer FDK AAC Codec Library for Android
+
+© Copyright  1995 - 2018 Fraunhofer-Gesellschaft zur Förderung der angewandten
+Forschung e.V. All rights reserved.
+
+ 1.    INTRODUCTION
+The Fraunhofer FDK AAC Codec Library for Android ("FDK AAC Codec") is software
+that implements the MPEG Advanced Audio Coding ("AAC") encoding and decoding
+scheme for digital audio. This FDK AAC Codec software is intended to be used on
+a wide variety of Android devices.
+
+AAC's HE-AAC and HE-AAC v2 versions are regarded as today's most efficient
+general perceptual audio codecs. AAC-ELD is considered the best-performing
+full-bandwidth communications codec by independent studies and is widely
+deployed. AAC has been standardized by ISO and IEC as part of the MPEG
+specifications.
+
+Patent licenses for necessary patent claims for the FDK AAC Codec (including
+those of Fraunhofer) may be obtained through Via Licensing
+(www.vialicensing.com) or through the respective patent owners individually for
+the purpose of encoding or decoding bit streams in products that are compliant
+with the ISO/IEC MPEG audio standards. Please note that most manufacturers of
+Android devices already license these patent claims through Via Licensing or
+directly from the patent owners, and therefore FDK AAC Codec software may
+already be covered under those patent licenses when it is used for those
+licensed purposes only.
+
+Commercially-licensed AAC software libraries, including floating-point versions
+with enhanced sound quality, are also available from Fraunhofer. Users are
+encouraged to check the Fraunhofer website for additional applications
+information and documentation.
+
+2.    COPYRIGHT LICENSE
+
+Redistribution and use in source and binary forms, with or without modification,
+are permitted without payment of copyright license fees provided that you
+satisfy the following conditions:
+
+You must retain the complete text of this software license in redistributions of
+the FDK AAC Codec or your modifications thereto in source code form.
+
+You must retain the complete text of this software license in the documentation
+and/or other materials provided with redistributions of the FDK AAC Codec or
+your modifications thereto in binary form. You must make available free of
+charge copies of the complete source code of the FDK AAC Codec and your
+modifications thereto to recipients of copies in binary form.
+
+The name of Fraunhofer may not be used to endorse or promote products derived
+from this library without prior written permission.
+
+You may not charge copyright license fees for anyone to use, copy or distribute
+the FDK AAC Codec software or your modifications thereto.
+
+Your modified versions of the FDK AAC Codec must carry prominent notices stating
+that you changed the software and the date of any change. For modified versions
+of the FDK AAC Codec, the term "Fraunhofer FDK AAC Codec Library for Android"
+must be replaced by the term "Third-Party Modified Version of the Fraunhofer FDK
+AAC Codec Library for Android."
+
+3.    NO PATENT LICENSE
+
+NO EXPRESS OR IMPLIED LICENSES TO ANY PATENT CLAIMS, including without
+limitation the patents of Fraunhofer, ARE GRANTED BY THIS SOFTWARE LICENSE.
+Fraunhofer provides no warranty of patent non-infringement with respect to this
+software.
+
+You may use this FDK AAC Codec software or modifications thereto only for
+purposes that are authorized by appropriate patent licenses.
+
+4.    DISCLAIMER
+
+This FDK AAC Codec software is provided by Fraunhofer on behalf of the copyright
+holders and contributors "AS IS" and WITHOUT ANY EXPRESS OR IMPLIED WARRANTIES,
+including but not limited to the implied warranties of merchantability and
+fitness for a particular purpose. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+CONTRIBUTORS BE LIABLE for any direct, indirect, incidental, special, exemplary,
+or consequential damages, including but not limited to procurement of substitute
+goods or services; loss of use, data, or profits, or business interruption,
+however caused and on any theory of liability, whether in contract, strict
+liability, or tort (including negligence), arising in any way out of the use of
+this software, even if advised of the possibility of such damage.
+
+5.    CONTACT INFORMATION
+
+Fraunhofer Institute for Integrated Circuits IIS
+Attention: Audio and Multimedia Departments - FDK AAC LL
+Am Wolfsmantel 33
+91058 Erlangen, Germany
+
+www.iis.fraunhofer.de/amm
+amm-info@iis.fraunhofer.de
+----------------------------------------------------------------------------- */
+
+/******************* Library for basic calculation routines ********************
+
+   Author(s):   Markus Lohwasser, Josef Hoepfl, Manuel Jander
+
+   Description: QMF filterbank
+
+*******************************************************************************/
+
+#ifndef QMF_PCM_H
+#define QMF_PCM_H
+
+/*
+   All Synthesis functions dependent on datatype INT_PCM_QMFOUT
+   Should only be included by qmf.cpp, but not compiled separately, please
+   exclude compilation from project, if done otherwise. Is optional included
+   twice to duplicate all functions with two different pre-definitions, as:
+        #define INT_PCM_QMFOUT LONG
+    and ...
+        #define INT_PCM_QMFOUT SHORT
+    needed to run QMF synthesis in both 16bit and 32bit sample output format.
+*/
+
+#define QSSCALE (0)
+#define FX_DBL2FX_QSS(x) (x)
+#define FX_QSS2FX_DBL(x) (x)
+
+/*!
+  \brief Perform Synthesis Prototype Filtering on a single slot of input data.
+
+  The filter takes 2 * qmf->no_channels of input data and
+  generates qmf->no_channels time domain output samples.
+*/
+/* static */
+#ifndef FUNCTION_qmfSynPrototypeFirSlot
+void qmfSynPrototypeFirSlot(
+#else
+void qmfSynPrototypeFirSlot_fallback(
+#endif
+    HANDLE_QMF_FILTER_BANK qmf,
+    FIXP_DBL *RESTRICT realSlot,      /*!< Input: Pointer to real Slot */
+    FIXP_DBL *RESTRICT imagSlot,      /*!< Input: Pointer to imag Slot */
+    INT_PCM_QMFOUT *RESTRICT timeOut, /*!< Time domain data */
+    int stride) {
+  FIXP_QSS *FilterStates = (FIXP_QSS *)qmf->FilterStates;
+  int no_channels = qmf->no_channels;
+  const FIXP_PFT *p_Filter = qmf->p_filter;
+  int p_stride = qmf->p_stride;
+  int j;
+  FIXP_QSS *RESTRICT sta = FilterStates;
+  const FIXP_PFT *RESTRICT p_flt, *RESTRICT p_fltm;
+  int scale = (DFRACT_BITS - SAMPLE_BITS_QMFOUT) - 1 - qmf->outScalefactor -
+              qmf->outGain_e;
+
+  p_flt =
+      p_Filter + p_stride * QMF_NO_POLY; /*                     5th of 330 */
+  p_fltm = p_Filter + (qmf->FilterSize / 2) -
+           p_stride * QMF_NO_POLY; /* 5 + (320 - 2*5) = 315th of 330 */
+
+  FIXP_SGL gain = FX_DBL2FX_SGL(qmf->outGain_m);
+
+  FIXP_DBL rnd_val = 0;
+
+  if (scale > 0) {
+    if (scale < (DFRACT_BITS - 1))
+      rnd_val = FIXP_DBL(1 << (scale - 1));
+    else
+      scale = (DFRACT_BITS - 1);
+  } else {
+    scale = fMax(scale, -(DFRACT_BITS - 1));
+  }
+
+  for (j = no_channels - 1; j >= 0; j--) {
+    FIXP_DBL imag = imagSlot[j]; /* no_channels-1 .. 0 */
+    FIXP_DBL real = realSlot[j]; /* no_channels-1 .. 0 */
+    {
+      INT_PCM_QMFOUT tmp;
+      FIXP_DBL Are = fMultAddDiv2(FX_QSS2FX_DBL(sta[0]), p_fltm[0], real);
+
+      /* This PCM formatting performs:
+         - multiplication with 16-bit gain, if not -1.0f
+         - rounding, if shift right is applied
+         - apply shift left (or right) with saturation to 32 (or 16) bits
+         - store output with --stride in 32 (or 16) bit format
+      */
+      if (gain != (FIXP_SGL)(-32768)) /* -1.0f */
+      {
+        Are = fMult(Are, gain);
+      }
+      if (scale >= 0) {
+        FDK_ASSERT(
+            Are <=
+            (Are + rnd_val)); /* Round-addition must not overflow, might be
+                                 equal for rnd_val=0 */
+        tmp = (INT_PCM_QMFOUT)(
+            SATURATE_RIGHT_SHIFT(Are + rnd_val, scale, SAMPLE_BITS_QMFOUT));
+      } else {
+        tmp = (INT_PCM_QMFOUT)(
+            SATURATE_LEFT_SHIFT(Are, -scale, SAMPLE_BITS_QMFOUT));
+      }
+
+      { timeOut[(j)*stride] = tmp; }
+    }
+
+    sta[0] = FX_DBL2FX_QSS(fMultAddDiv2(FX_QSS2FX_DBL(sta[1]), p_flt[4], imag));
+    sta[1] =
+        FX_DBL2FX_QSS(fMultAddDiv2(FX_QSS2FX_DBL(sta[2]), p_fltm[1], real));
+    sta[2] = FX_DBL2FX_QSS(fMultAddDiv2(FX_QSS2FX_DBL(sta[3]), p_flt[3], imag));
+    sta[3] =
+        FX_DBL2FX_QSS(fMultAddDiv2(FX_QSS2FX_DBL(sta[4]), p_fltm[2], real));
+    sta[4] = FX_DBL2FX_QSS(fMultAddDiv2(FX_QSS2FX_DBL(sta[5]), p_flt[2], imag));
+    sta[5] =
+        FX_DBL2FX_QSS(fMultAddDiv2(FX_QSS2FX_DBL(sta[6]), p_fltm[3], real));
+    sta[6] = FX_DBL2FX_QSS(fMultAddDiv2(FX_QSS2FX_DBL(sta[7]), p_flt[1], imag));
+    sta[7] =
+        FX_DBL2FX_QSS(fMultAddDiv2(FX_QSS2FX_DBL(sta[8]), p_fltm[4], real));
+    sta[8] = FX_DBL2FX_QSS(fMultDiv2(p_flt[0], imag));
+    p_flt += (p_stride * QMF_NO_POLY);
+    p_fltm -= (p_stride * QMF_NO_POLY);
+    sta += 9;  // = (2*QMF_NO_POLY-1);
+  }
+}
+
+#ifndef FUNCTION_qmfSynPrototypeFirSlot_NonSymmetric
+/*!
+  \brief Perform Synthesis Prototype Filtering on a single slot of input data.
+
+  The filter takes 2 * qmf->no_channels of input data and
+  generates qmf->no_channels time domain output samples.
+*/
+static void qmfSynPrototypeFirSlot_NonSymmetric(
+    HANDLE_QMF_FILTER_BANK qmf,
+    FIXP_DBL *RESTRICT realSlot,      /*!< Input: Pointer to real Slot */
+    FIXP_DBL *RESTRICT imagSlot,      /*!< Input: Pointer to imag Slot */
+    INT_PCM_QMFOUT *RESTRICT timeOut, /*!< Time domain data */
+    int stride) {
+  FIXP_QSS *FilterStates = (FIXP_QSS *)qmf->FilterStates;
+  int no_channels = qmf->no_channels;
+  const FIXP_PFT *p_Filter = qmf->p_filter;
+  int p_stride = qmf->p_stride;
+  int j;
+  FIXP_QSS *RESTRICT sta = FilterStates;
+  const FIXP_PFT *RESTRICT p_flt, *RESTRICT p_fltm;
+  int scale = (DFRACT_BITS - SAMPLE_BITS_QMFOUT) - 1 - qmf->outScalefactor -
+              qmf->outGain_e;
+
+  p_flt = p_Filter; /*!< Pointer to first half of filter coefficients */
+  p_fltm =
+      &p_flt[qmf->FilterSize / 2]; /* at index 320, overall 640 coefficients */
+
+  FIXP_SGL gain = FX_DBL2FX_SGL(qmf->outGain_m);
+
+  FIXP_DBL rnd_val = (FIXP_DBL)0;
+
+  if (scale > 0) {
+    if (scale < (DFRACT_BITS - 1))
+      rnd_val = FIXP_DBL(1 << (scale - 1));
+    else
+      scale = (DFRACT_BITS - 1);
+  } else {
+    scale = fMax(scale, -(DFRACT_BITS - 1));
+  }
+
+  for (j = no_channels - 1; j >= 0; j--) {
+    FIXP_DBL imag = imagSlot[j]; /* no_channels-1 .. 0 */
+    FIXP_DBL real = realSlot[j]; /* no_channels-1 .. 0 */
+    {
+      INT_PCM_QMFOUT tmp;
+      FIXP_DBL Are = sta[0] + FX_DBL2FX_QSS(fMultDiv2(p_fltm[4], real));
+
+      /* This PCM formatting performs:
+         - multiplication with 16-bit gain, if not -1.0f
+         - rounding, if shift right is applied
+         - apply shift left (or right) with saturation to 32 (or 16) bits
+         - store output with --stride in 32 (or 16) bit format
+      */
+      if (gain != (FIXP_SGL)(-32768)) /* -1.0f */
+      {
+        Are = fMult(Are, gain);
+      }
+      if (scale > 0) {
+        FDK_ASSERT(Are <
+                   (Are + rnd_val)); /* Round-addition must not overflow */
+        tmp = (INT_PCM_QMFOUT)(
+            SATURATE_RIGHT_SHIFT(Are + rnd_val, scale, SAMPLE_BITS_QMFOUT));
+      } else {
+        tmp = (INT_PCM_QMFOUT)(
+            SATURATE_LEFT_SHIFT(Are, -scale, SAMPLE_BITS_QMFOUT));
+      }
+      timeOut[j * stride] = tmp;
+    }
+
+    sta[0] = sta[1] + FX_DBL2FX_QSS(fMultDiv2(p_flt[4], imag));
+    sta[1] = sta[2] + FX_DBL2FX_QSS(fMultDiv2(p_fltm[3], real));
+    sta[2] = sta[3] + FX_DBL2FX_QSS(fMultDiv2(p_flt[3], imag));
+
+    sta[3] = sta[4] + FX_DBL2FX_QSS(fMultDiv2(p_fltm[2], real));
+    sta[4] = sta[5] + FX_DBL2FX_QSS(fMultDiv2(p_flt[2], imag));
+    sta[5] = sta[6] + FX_DBL2FX_QSS(fMultDiv2(p_fltm[1], real));
+    sta[6] = sta[7] + FX_DBL2FX_QSS(fMultDiv2(p_flt[1], imag));
+
+    sta[7] = sta[8] + FX_DBL2FX_QSS(fMultDiv2(p_fltm[0], real));
+    sta[8] = FX_DBL2FX_QSS(fMultDiv2(p_flt[0], imag));
+
+    p_flt += (p_stride * QMF_NO_POLY);
+    p_fltm += (p_stride * QMF_NO_POLY);
+    sta += 9;  // = (2*QMF_NO_POLY-1);
+  }
+}
+#endif /* FUNCTION_qmfSynPrototypeFirSlot_NonSymmetric */
+
+void qmfSynthesisFilteringSlot(HANDLE_QMF_FILTER_BANK synQmf,
+                               const FIXP_DBL *realSlot,
+                               const FIXP_DBL *imagSlot,
+                               const int scaleFactorLowBand,
+                               const int scaleFactorHighBand,
+                               INT_PCM_QMFOUT *timeOut, const int stride,
+                               FIXP_DBL *pWorkBuffer) {
+  if (!(synQmf->flags & QMF_FLAG_LP))
+    qmfInverseModulationHQ(synQmf, realSlot, imagSlot, scaleFactorLowBand,
+                           scaleFactorHighBand, pWorkBuffer);
+  else {
+    if (synQmf->flags & QMF_FLAG_CLDFB) {
+      qmfInverseModulationLP_odd(synQmf, realSlot, scaleFactorLowBand,
+                                 scaleFactorHighBand, pWorkBuffer);
+    } else {
+      qmfInverseModulationLP_even(synQmf, realSlot, scaleFactorLowBand,
+                                  scaleFactorHighBand, pWorkBuffer);
+    }
+  }
+
+  if (synQmf->flags & QMF_FLAG_NONSYMMETRIC) {
+    qmfSynPrototypeFirSlot_NonSymmetric(synQmf, pWorkBuffer,
+                                        pWorkBuffer + synQmf->no_channels,
+                                        timeOut, stride);
+  } else {
+    qmfSynPrototypeFirSlot(synQmf, pWorkBuffer,
+                           pWorkBuffer + synQmf->no_channels, timeOut, stride);
+  }
+}
+
+/*!
+ *
+ * \brief Perform complex-valued subband synthesis of the
+ *        low band and the high band and store the
+ *        time domain data in timeOut
+ *
+ * First step: Calculate the proper scaling factor of current
+ * spectral data in qmfReal/qmfImag, old spectral data in the overlap
+ * range and filter states.
+ *
+ * Second step: Perform Frequency-to-Time mapping with inverse
+ * Modulation slot-wise.
+ *
+ * Third step: Perform FIR-filter slot-wise. To save space for filter
+ * states, the MAC operations are executed directly on the filter states
+ * instead of accumulating several products in the accumulator. The
+ * buffer shift at the end of the function should be replaced by a
+ * modulo operation, which is available on some DSPs.
+ *
+ * Last step: Copy the upper part of the spectral data to the overlap buffer.
+ *
+ * The qmf coefficient table is symmetric. The symmetry is exploited by
+ * shrinking the coefficient table to half the size. The addressing mode
+ * takes care of the symmetries.  If the #define #QMFTABLE_FULL is set,
+ * coefficient addressing works on the full table size. The code will be
+ * slightly faster and slightly more compact.
+ *
+ * Workbuffer requirement: 2 x sizeof(**QmfBufferReal) * synQmf->no_channels
+ * The workbuffer must be aligned
+ */
+void qmfSynthesisFiltering(
+    HANDLE_QMF_FILTER_BANK synQmf, /*!< Handle of Qmf Synthesis Bank  */
+    FIXP_DBL **QmfBufferReal,      /*!< Low and High band, real */
+    FIXP_DBL **QmfBufferImag,      /*!< Low and High band, imag */
+    const QMF_SCALE_FACTOR *scaleFactor,
+    const INT ov_len,        /*!< split Slot of overlap and actual slots */
+    INT_PCM_QMFOUT *timeOut, /*!< Pointer to output */
+    const INT stride,        /*!< stride factor of output */
+    FIXP_DBL *pWorkBuffer    /*!< pointer to temporal working buffer */
+) {
+  int i;
+  int L = synQmf->no_channels;
+  int scaleFactorHighBand;
+  int scaleFactorLowBand_ov, scaleFactorLowBand_no_ov;
+
+  FDK_ASSERT(synQmf->no_channels >= synQmf->lsb);
+  FDK_ASSERT(synQmf->no_channels >= synQmf->usb);
+
+  /* adapt scaling */
+  scaleFactorHighBand = -ALGORITHMIC_SCALING_IN_ANALYSIS_FILTERBANK -
+                        scaleFactor->hb_scale - synQmf->filterScale;
+  scaleFactorLowBand_ov = -ALGORITHMIC_SCALING_IN_ANALYSIS_FILTERBANK -
+                          scaleFactor->ov_lb_scale - synQmf->filterScale;
+  scaleFactorLowBand_no_ov = -ALGORITHMIC_SCALING_IN_ANALYSIS_FILTERBANK -
+                             scaleFactor->lb_scale - synQmf->filterScale;
+
+  for (i = 0; i < synQmf->no_col; i++) /* ----- no_col loop ----- */
+  {
+    const FIXP_DBL *QmfBufferImagSlot = NULL;
+
+    int scaleFactorLowBand =
+        (i < ov_len) ? scaleFactorLowBand_ov : scaleFactorLowBand_no_ov;
+
+    if (!(synQmf->flags & QMF_FLAG_LP)) QmfBufferImagSlot = QmfBufferImag[i];
+
+    qmfSynthesisFilteringSlot(synQmf, QmfBufferReal[i], QmfBufferImagSlot,
+                              scaleFactorLowBand, scaleFactorHighBand,
+                              timeOut + (i * L * stride), stride, pWorkBuffer);
+  } /* no_col loop  i  */
+}
+#endif /* QMF_PCM_H */