Line data    Source code

       1             : /******************************************************************************************************
       2             : 
       3             :    (C) 2022-2025 IVAS codec Public Collaboration with portions copyright Dolby International AB, Ericsson AB,
       4             :    Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V., Huawei Technologies Co. LTD.,
       5             :    Koninklijke Philips N.V., Nippon Telegraph and Telephone Corporation, Nokia Technologies Oy, Orange,
       6             :    Panasonic Holdings Corporation, Qualcomm Technologies, Inc., VoiceAge Corporation, and other
       7             :    contributors to this repository. All Rights Reserved.
       8             : 
       9             :    This software is protected by copyright law and by international treaties.
      10             :    The IVAS codec Public Collaboration consisting of Dolby International AB, Ericsson AB,
      11             :    Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V., Huawei Technologies Co. LTD.,
      12             :    Koninklijke Philips N.V., Nippon Telegraph and Telephone Corporation, Nokia Technologies Oy, Orange,
      13             :    Panasonic Holdings Corporation, Qualcomm Technologies, Inc., VoiceAge Corporation, and other
      14             :    contributors to this repository retain full ownership rights in their respective contributions in
      15             :    the software. This notice grants no license of any kind, including but not limited to patent
      16             :    license, nor is any license granted by implication, estoppel or otherwise.
      17             : 
      18             :    Contributors are required to enter into the IVAS codec Public Collaboration agreement before making
      19             :    contributions.
      20             : 
      21             :    This software is provided "AS IS", without any express or implied warranties. The software is in the
      22             :    development stage. It is intended exclusively for experts who have experience with such software and
      23             :    solely for the purpose of inspection. All implied warranties of non-infringement, merchantability
      24             :    and fitness for a particular purpose are hereby disclaimed and excluded.
      25             : 
      26             :    Any dispute, controversy or claim arising under or in relation to providing this software shall be
      27             :    submitted to and settled by the final, binding jurisdiction of the courts of Munich, Germany in
      28             :    accordance with the laws of the Federal Republic of Germany excluding its conflict of law rules and
      29             :    the United Nations Convention on Contracts on the International Sales of Goods.
      30             : 
      31             : *******************************************************************************************************/
      32             : 
      33             : /*====================================================================================
      34             :     EVS Codec 3GPP TS26.443 Nov 04, 2021. Version 12.14.0 / 13.10.0 / 14.6.0 / 15.4.0 / 16.3.0
      35             :   ====================================================================================*/
      36             : 
      37             : #include <stdint.h>
      38             : #include "options.h"
      39             : #ifdef DEBUGGING
      40             : #include "debug.h"
      41             : #endif
      42             : #include <math.h>
      43             : #include "cnst.h"
      44             : #include "prot.h"
      45             : #include "wmc_auto.h"
      46             : 
      47             : /*----------------------------------------------------------------------------------*
      48             :  * fer_energy()
      49             :  *
      50             :  * Estimation of pitch-synchronous (voiced sounds) or half-frame energy
      51             :  *----------------------------------------------------------------------------------*/
      52             : 
      53      874150 : void fer_energy(
      54             :     const int16_t L_frame, /* i  : frame length                           */
      55             :     const int16_t clas,    /* i  : frame classification                   */
      56             :     const float *synth,    /* i  : synthesized speech at Fs = 12k8 Hz     */
      57             :     const float pitch,     /* i  : pitch period                           */
      58             :     float *enr,            /* o  : pitch-synchronous or half_frame energy */
      59             :     const int16_t offset   /* i  : speech pointer offset (0 or L_frame)   */
      60             : )
      61             : {
      62             :     int16_t len;
      63             :     const float *pt_synth;
      64             : 
      65      874150 :     if ( clas == VOICED_CLAS || clas == ONSET || clas == SIN_ONSET ) /* Voiced or Onset current frame */
      66             :     {
      67      333922 :         len = (int16_t) ( pitch + 0.5f ); /* pitch value */
      68             : 
      69      333922 :         pt_synth = synth;
      70      333922 :         if ( offset != 0 )
      71             :         {
      72      329737 :             pt_synth = synth + L_frame - len;
      73             :         }
      74             : 
      75      333922 :         emaximum( pt_synth, len, enr ); /* pitch synchronous E */
      76             :     }
      77             :     else
      78             :     {
      79      540228 :         pt_synth = synth;
      80      540228 :         if ( offset != 0 )
      81             :         {
      82      538515 :             pt_synth = synth + L_frame / 2;
      83             :         }
      84             : 
      85      540228 :         *enr = dotp( pt_synth, pt_synth, L_frame / 2 );
      86      540228 :         *enr /= (float) ( L_frame / 2 );
      87             :     }
      88      874150 :     return;
      89             : }
      90             : 
      91             : 
      92             : /*------------------------------------------------------------------------*
      93             :  * frame_energy()
      94             :  *
      95             :  * Compute pitch-synchronous energy at the frame end
      96             :  *------------------------------------------------------------------------*/
      97             : 
      98      411723 : float frame_energy(
      99             :     const int16_t L_frame, /* i  : length of the frame                            */
     100             :     const float *pitch,    /* i  : pitch values for each subframe                 */
     101             :     const float *speech,   /* i  : pointer to speech signal for E computation     */
     102             :     const float lp_speech, /* i  : long-term active speech energy average         */
     103             :     float *frame_ener      /* o  : pitch-synchronous energy at frame end          */
     104             : )
     105             : {
     106             :     float enern;
     107             :     const float *pt1;
     108             :     int16_t len;
     109             :     float dotProd;
     110             : 
     111      411723 :     len = (int16_t) ( 0.5f * ( pitch[2] + pitch[3] ) + 0.5f );
     112      411723 :     if ( len < L_SUBFR )
     113             :     {
     114      102855 :         len *= 2;
     115             :     }
     116             : 
     117      411723 :     pt1 = speech + L_frame - len;
     118             : 
     119      411723 :     dotProd = dotp( pt1, pt1, len );
     120      411723 :     if ( 0 == dotProd )
     121             :     {
     122          66 :         *frame_ener = MIN_LOG_VAL_60dB;
     123             :     }
     124             :     else
     125             :     {
     126      411657 :         *frame_ener = 10.0f * (float) log10( dotProd / (float) len );
     127             :     }
     128      411723 :     enern = *frame_ener - lp_speech;
     129             : 
     130      411723 :     return enern;
     131             : }