autocorr.cpp

实现3GPP的GSM中AMR语音的CODECS。

/* ------------------------------------------------------------------
 * Copyright (C) 2008 PacketVideo
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *      http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either
 * express or implied.
 * See the License for the specific language governing permissions
 * and limitations under the License.
 * -------------------------------------------------------------------
 */
/****************************************************************************************
Portions of this file are derived from the following 3GPP standard:

    3GPP TS 26.073
    ANSI-C code for the Adaptive Multi-Rate (AMR) speech codec
    Available from http://www.3gpp.org

(C) 2004, 3GPP Organizational Partners (ARIB, ATIS, CCSA, ETSI, TTA, TTC)
Permission to distribute, modify and use this file under the standard license
terms listed above has been obtained from the copyright holder.
****************************************************************************************/
/*
------------------------------------------------------------------------------



 Pathname: ./audio/gsm-amr/c/src/autocorr.c

     Date: 05/15/2000

------------------------------------------------------------------------------
 REVISION HISTORY

 Description: Put into template...starting optimization.

 Description: Removed call to mult_r routine.

 Description: Modified Input/Output Definitions section to comply with the
          current template. Fixed tabs.

 Description: Updated Input/Output definitions by making them more
          descriptive.

 Description: Synchronized file with UMTS version 3.2.0. Updated coding
              template.

 Description: Made the following changes per comments from Phase 2/3 review:
              1. Added full pathname of file.
              2. Fixed typecasting issue with TI compiler.
              3. Modified FOR loops to count down.
              4. Added comment to the code.

 Description: Removed extern to global paramter (Flag Overflow) and replaced
 by passing in a pointer to Overflow.  Also, made several small changes to
 bring code more in line with PV Standards.

 Description:
            1. Added pointer to avoid adding offsets in every pass
            2. Break last loop in two nested loop to speed up processing
            3. Removed extra check for overflow by doing scaling right
               after overflow is detected.
            4. Eliminated calls to basic operations (like extract) not
               needed because of the nature of the number (all bounded)

 Description:
              1. Fixed for:
                overflow check was looking for positive number before a left
                shift. When numbers were big enough, positive numbers after
                shifted became negative, causing a 1/0 division).
                Fixed so now it checks for numbers lesser than 0x40000000
                before the left shift

 Description:
              1.Modified check for saturation to match bit exact test.
                Also, when saturation is reached, a faster loop is used
                (with no energy accumulation) to speed up processing


 Description:
              1.Added pointer initialization to for loop when saturation
                is found. This because some compiler ( like Vcpp in release
                mode) when optimizing code, may remove pointer information
                once the loop is broken.

 Description:  Added casting to eliminate warnings

 Description:  Replaced "int" and/or "char" with OSCL defined types.

 Description: Using inlines from fxp_arithmetic.h.

 Description: Replacing fxp_arithmetic.h with basic_op.h.

 Description:

----------------------------------------------------------------------------*/

/*----------------------------------------------------------------------------
; INCLUDES
----------------------------------------------------------------------------*/
#include "autocorr.h"
#include "typedef.h"
#include "basic_op.h"
#include "oper_32b.h"
#include "cnst.h"

/*----------------------------------------------------------------------------
; MACROS
; Define module specific macros here
----------------------------------------------------------------------------*/


/*----------------------------------------------------------------------------
; DEFINES
; Include all pre-processor statements here. Include conditional
; compile variables also.
----------------------------------------------------------------------------*/


/*----------------------------------------------------------------------------
; LOCAL FUNCTION DEFINITIONS
; Function Prototype declaration
----------------------------------------------------------------------------*/

/*----------------------------------------------------------------------------
; LOCAL STORE/BUFFER/POINTER DEFINITIONS
; Variable declaration - defined here and used outside this module
----------------------------------------------------------------------------*/

/*----------------------------------------------------------------------------
; EXTERNAL FUNCTION REFERENCES
; Declare functions defined elsewhere and referenced in this module
----------------------------------------------------------------------------*/


/*----------------------------------------------------------------------------
; EXTERNAL GLOBAL STORE/BUFFER/POINTER REFERENCES
; Declare variables used in this module but defined elsewhere
----------------------------------------------------------------------------*/

/*
------------------------------------------------------------------------------
 FUNCTION NAME: Autocorr
----------------------------------------------------------------------------
 INPUT AND OUTPUT DEFINITIONS

 Inputs:
    x = buffer of input signals of type Word16
    m = LPC order of type Word16
    wind = buffer of window signals of type Word16
    r_h = buffer containing the high word of the autocorrelation values
          of type Word16
    r_l = buffer containing the low word of the autocorrelation values
          of type Word16

    pOverflow = pointer to variable of type Flag *, which indicates if
                overflow occurs.

 Outputs:
    r_h buffer contains the high word of the new autocorrelation values
    r_l buffer contains the low word of the new autocorrelation values
    pOverflow -> 1 if overflow occurs.

 Returns:
    norm = normalized autocorrelation at lag zero of type Word16

 Global Variables Used:
    None

 Local Variables Needed:
    None

------------------------------------------------------------------------------
 FUNCTION DESCRIPTION

 This function windows the input signal with the provided window
 then calculates the autocorrelation values for lags of 0,1,...m,
 where m is the passed in LPC order.

------------------------------------------------------------------------------
 REQUIREMENTS

 None.

------------------------------------------------------------------------------
 REFERENCES

 autocorr.c, UMTS GSM AMR speech codec, R99 - Version 3.2.0, March 2, 2001

------------------------------------------------------------------------------
 PSEUDO-CODE

Word16 Autocorr (
    Word16 x[],            // (i)    : Input signal (L_WINDOW)
    Word16 m,              // (i)    : LPC order
    Word16 r_h[],          // (o)    : Autocorrelations  (msb)
    Word16 r_l[],          // (o)    : Autocorrelations  (lsb)
    const Word16 wind[]    // (i)    : window for LPC analysis (L_WINDOW)
)
{
    Word16 i, j, norm;
    Word16 y[L_WINDOW];
    Word32 sum;
    Word16 overfl, overfl_shft;

    // Windowing of signal

    for (i = 0; i < L_WINDOW; i++)
    {
        y[i] = mult_r (x[i], wind[i]);
    }

    // Compute r[0] and test for overflow

    overfl_shft = 0;

    do
    {
        overfl = 0;
        sum = 0L;

        for (i = 0; i < L_WINDOW; i++)
        {
            sum = L_mac (sum, y[i], y[i]);
        }

        // If overflow divide y[] by 4

        if (L_sub (sum, MAX_32) == 0L)
        {
            overfl_shft = add (overfl_shft, 4);
            overfl = 1; // Set the overflow flag

            for (i = 0; i < L_WINDOW; i++)
            {
                y[i] = shr (y[i], 2);
            }
        }
    }
    while (overfl != 0);

    sum = L_add (sum, 1L);             // Avoid the case of all zeros

    // Normalization of r[0]

    norm = norm_l (sum);
    sum = L_shl (sum, norm);
    L_Extract (sum, &r_h[0], &r_l[0]); // Put in DPF format (see oper_32b)

    // r[1] to r[m]

    for (i = 1; i <= m; i++)
    {
        sum = 0;

        for (j = 0; j < L_WINDOW - i; j++)
        {
            sum = L_mac (sum, y[j], y[j + i]);
        }

        sum = L_shl (sum, norm);
        L_Extract (sum, &r_h[i], &r_l[i]);
    }

    norm = sub (norm, overfl_shft);

    return norm;
}


------------------------------------------------------------------------------
 RESOURCES USED [optional]

 When the code is written for a specific target processor the
 the resources used should be documented below.

 HEAP MEMORY USED: x bytes

 STACK MEMORY USED: x bytes

 CLOCK CYCLES: (cycle count equation for this function) + (variable
                used to represent cycle count for each subroutine
                called)
     where: (cycle count variable) = cycle count for [subroutine
                                     name]

------------------------------------------------------------------------------
 CAUTION [optional]
 [State any special notes, constraints or cautions for users of this function]

------------------------------------------------------------------------------
*/

Word16 Autocorr(
    Word16 x[],            /* (i)    : Input signal (L_WINDOW)            */
    Word16 m,              /* (i)    : LPC order                          */
    Word16 r_h[],          /* (o)    : Autocorrelations  (msb)            */
    Word16 r_l[],          /* (o)    : Autocorrelations  (lsb)            */
    const Word16 wind[],   /* (i)    : window for LPC analysis (L_WINDOW) */
    Flag  *pOverflow       /* (o)    : indicates overflow                 */
)
{
    register Word16 i;
    register Word16 j;
    register Word16 norm;

    Word16 y[L_WINDOW];
    Word32 sum;
    Word16 overfl_shft;


    /* Added for optimization  */


    Word16 temp;
    Word16 *p_x;
    Word16 *p_y;
    Word16 *p_y_1;
    Word16 *p_y_ref;
    Word16 *p_rh;
    Word16 *p_rl;
    const Word16 *p_wind;
    p_y = y;
    p_x = x;
    p_wind = wind;
    /*
     *  Windowing of the signal
     */

    OSCL_UNUSED_ARG(pOverflow);

    sum = 0L;
    j = 0;

    for (i = L_WINDOW; i != 0; i--)
    {
        temp = (amrnb_fxp_mac_16_by_16bb((Word32) * (p_x++), (Word32) * (p_wind++), 0x04000)) >> 15;
        *(p_y++) = temp;

        sum += ((Word32)temp * temp) << 1;
        if (sum < 0)
        {
            /*
             * if oveflow exist, then stop accumulation
             */
            j = 1;
            break;
        }

    }
    /*
     * if oveflow existed, complete  windowing operation
     * without computing energy
     */

    if (j)
    {
        p_y = &y[L_WINDOW-i];
        p_x = &x[L_WINDOW-i];
        p_wind = &wind[L_WINDOW-i];

        for (; i != 0; i--)
        {
            temp = (amrnb_fxp_mac_16_by_16bb((Word32) * (p_x++), (Word32) * (p_wind++), 0x04000)) >> 15;
            *(p_y++) = temp;
        }
    }


    /*
     *  Compute r[0] and test for overflow
     */

    overfl_shft = 0;

    /*
     * scale down by 1/4 only when needed
     */
    while (j == 1)
    {
        /* If overflow divide y[] by 4          */
        /* FYI: For better resolution, we could */
        /*      divide y[] by 2                 */
        overfl_shft += 4;
        p_y   = &y[0];
        sum = 0L;

        for (i = (L_WINDOW >> 1); i != 0 ; i--)
        {
            temp = *p_y >> 2;
            *(p_y++) = temp;
            sum += ((Word32)temp * temp) << 1;
            temp = *p_y >> 2;
            *(p_y++) = temp;
            sum += ((Word32)temp * temp) << 1;
        }
        if (sum > 0)
        {
            j = 0;
        }

    }

    sum += 1L;              /* Avoid the case of all zeros */

    /* Normalization of r[0] */

    norm = norm_l(sum);

    sum <<= norm;

    /* Put in DPF format (see oper_32b) */
    r_h[0] = (Word16)(sum >> 16);
    r_l[0] = (Word16)((sum >> 1) - ((Word32)(r_h[0]) << 15));

    /* r[1] to r[m] */

    p_y_ref = &y[L_WINDOW - 1 ];
    p_rh = &r_h[m];
    p_rl = &r_l[m];

    for (i = m; i > 0; i--)
    {
        sum  = 0;

        p_y   = &y[L_WINDOW - i - 1];
        p_y_1 = p_y_ref;

        for (j = (L_WINDOW - i - 1) >> 1; j != 0; j--)
        {
            sum = amrnb_fxp_mac_16_by_16bb((Word32) * (p_y--), (Word32) * (p_y_1--), sum);
            sum = amrnb_fxp_mac_16_by_16bb((Word32) * (p_y--), (Word32) * (p_y_1--), sum);
        }

        sum = amrnb_fxp_mac_16_by_16bb((Word32) * (p_y--), (Word32) * (p_y_1--), sum);

        if (((L_WINDOW - i - 1) & 1))
        {
            sum = amrnb_fxp_mac_16_by_16bb((Word32) * (p_y--), (Word32) * (p_y_1--), sum);
        }

        sum  <<= (norm + 1);

        *(p_rh)   = (Word16)(sum >> 16);
        *(p_rl--) = (Word16)((sum >> 1) - ((Word32) * (p_rh--) << 15));

    }

    norm -= overfl_shft;

    return (norm);

} /* Autocorr */