view libtwamr/sqrt_l.c @ 581:e2d5cad04cbf

libgsmhr1 RxFE: store CN R0+LPC separately from speech In the original GSM 06.06 code the ECU for speech mode is entirely separate from the CN generator, maintaining separate state. (The main intertie between them is the speech vs CN state variable, distinguishing between speech and CN BFIs, in addition to the CN-specific function of distinguishing between initial and update SIDs.) In the present RxFE implementation I initially thought that we could use the same saved_frame buffer for both ECU and CN, overwriting just the first 4 params (R0 and LPC) when a valid SID comes in. However, I now realize it was a bad idea: the original code has a corner case (long sequence of speech-mode BFIs to put the ECU in state 6, then SID and CN-mode BFIs, then a good speech frame) that would be broken by that buffer reuse approach. We could eliminate this corner case by resetting the ECU state when passing through a CN insertion period, but doing so would needlessly increase the behavioral diffs between GSM 06.06 and our version. Solution: use a separate CN-specific buffer for CN R0+LPC parameters, and match the behavior of GSM 06.06 code in this regard.
author Mychaela Falconia <falcon@freecalypso.org>
date Thu, 13 Feb 2025 10:02:45 +0000
parents 8dfb7cbe6b59
children
line wrap: on
line source

/*
********************************************************************************
*
*      GSM AMR-NB speech codec   R98   Version 7.6.0   December 12, 2001
*                                R99   Version 3.3.0                
*                                REL-4 Version 4.1.0                
*
********************************************************************************
*
*      File             : sqrt_l.c
*      Purpose          : Computes sqrt(L_x),  where  L_x is positive.
*                       : If L_var is negative or zero, the result is 0
*      Description      :
*            The function sqrt(L_x) is approximated by a table and linear
*            interpolation. The square root is computed using the
*            following steps:
*                1- Normalization of L_x.
*                2- If exponent is even then shift right once.
*                3- exponent = exponent/2
*                4- i = bit25-b31 of L_x;  16<=i<=63  because of normalization.
*                5- a = bit10-b24
*                6- i -=16
*                7- L_y = table[i]<<16 - (table[i] - table[i+1]) * a * 2
*                8- return L_y and exponent so caller can do denormalization
*
********************************************************************************
*/
/*
********************************************************************************
*                         MODULE INCLUDE FILE AND VERSION ID
********************************************************************************
*/
#include "namespace.h"
#include "sqrt_l.h"

/*
********************************************************************************
*                         INCLUDE FILES
********************************************************************************
*/
#include "typedef.h"
#include "basic_op.h"
#include "no_count.h"

/*
********************************************************************************
*                         LOCAL VARIABLES AND TABLES
********************************************************************************
*/
#include "sqrt_l.tab" /* Table for sqrt_l_exp() */

/*
********************************************************************************
*                         PUBLIC PROGRAM CODE
********************************************************************************
*/

Word32 sqrt_l_exp (/* o : output value,                          Q31 */
    Word32 L_x,    /* i : input value,                           Q31 */
    Word16 *exp    /* o : right shift to be applied to result,   Q1  */
)
{
    /*
          y = sqrt(x)

          x = f * 2^-e,   0.5 <= f < 1   (normalization)

          y = sqrt(f) * 2^(-e/2)  

          a) e = 2k   --> y = sqrt(f)   * 2^-k  (k = e div 2,
                                                 0.707 <= sqrt(f) < 1)
          b) e = 2k+1 --> y = sqrt(f/2) * 2^-k  (k = e div 2,
                                                 0.5 <= sqrt(f/2) < 0.707)
     */
    
    
    Word16 e, i, a, tmp;
    Word32 L_y;

    test (); 
    if (L_x <= (Word32) 0)
    {
        *exp = 0;               move16 ();
        return (Word32) 0;
    }

    e = norm_l (L_x) & 0xFFFE;  logic16 (); /* get next lower EVEN norm. exp  */
    L_x = L_shl (L_x, e);                   /* L_x is normalized to [0.25..1) */
    *exp = e;                   move16 ();  /* return 2*exponent (or Q1)      */

    L_x = L_shr (L_x, 9);
    i = extract_h (L_x);            /* Extract b25-b31, 16 <= i <= 63 because
                                       of normalization                       */
    L_x = L_shr (L_x, 1);   
    a = extract_l (L_x);            /* Extract b10-b24                        */
    a = a & (Word16) 0x7fff;    logic16 (); 

    i = sub (i, 16);                /* 0 <= i <= 47                           */

    L_y = L_deposit_h (table[i]);           /* table[i] << 16                 */
    tmp = sub (table[i], table[i + 1]);     /* table[i] - table[i+1])         */
    L_y = L_msu (L_y, tmp, a);              /* L_y -= tmp*a*2                 */
       
    /* L_y = L_shr (L_y, *exp); */          /* denormalization done by caller */

    return (L_y);
}