FreeCalypso > hg > gsm-codec-lib
view libtwamr/pitch_fr.c @ 507:06578223a752
libgsmhr1: integrate sp_rom.c
author | Mychaela Falconia <falcon@freecalypso.org> |
---|---|
date | Wed, 19 Jun 2024 02:13:56 +0000 |
parents | be8edf9e6bc1 |
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 : pitch_fr.c * Purpose : Find the pitch period with 1/3 or 1/6 subsample * : resolution (closed loop). * ******************************************************************************** */ /* ******************************************************************************** * MODULE INCLUDE FILE AND VERSION ID ******************************************************************************** */ #include "namespace.h" #include "pitch_fr.h" /* ******************************************************************************** * INCLUDE FILES ******************************************************************************** */ #include "typedef.h" #include "basic_op.h" #include "oper_32b.h" #include "no_count.h" #include "cnst.h" #include "enc_lag3.h" #include "enc_lag6.h" #include "inter_36.h" #include "inv_sqrt.h" #include "convolve.h" /* ******************************************************************************** * LOCAL VARIABLES AND TABLES ******************************************************************************** */ /* * mode dependent parameters used in Pitch_fr() * Note: order of MRxx in 'enum Mode' is important! */ static const struct { Word16 max_frac_lag; /* lag up to which fractional lags are used */ Word16 flag3; /* enable 1/3 instead of 1/6 fract. resolution */ Word16 first_frac; /* first fractional to check */ Word16 last_frac; /* last fractional to check */ Word16 delta_int_low; /* integer lag below TO to start search from */ Word16 delta_int_range; /* integer range around T0 */ Word16 delta_frc_low; /* fractional below T0 */ Word16 delta_frc_range; /* fractional range around T0 */ Word16 pit_min; /* minimum pitch */ } mode_dep_parm[8] = { /* MR475 */ { 84, 1, -2, 2, 5, 10, 5, 9, PIT_MIN }, /* MR515 */ { 84, 1, -2, 2, 5, 10, 5, 9, PIT_MIN }, /* MR59 */ { 84, 1, -2, 2, 3, 6, 5, 9, PIT_MIN }, /* MR67 */ { 84, 1, -2, 2, 3, 6, 5, 9, PIT_MIN }, /* MR74 */ { 84, 1, -2, 2, 3, 6, 5, 9, PIT_MIN }, /* MR795 */ { 84, 1, -2, 2, 3, 6, 10, 19, PIT_MIN }, /* MR102 */ { 84, 1, -2, 2, 3, 6, 5, 9, PIT_MIN }, /* MR122 */ { 94, 0, -3, 3, 3, 6, 5, 9, PIT_MIN_MR122 } }; /* ******************************************************************************** * LOCAL PROGRAM CODE ******************************************************************************** */ /************************************************************************* * * FUNCTION: Norm_Corr() * * PURPOSE: Find the normalized correlation between the target vector * and the filtered past excitation. * * DESCRIPTION: * The normalized correlation is given by the correlation between the * target and filtered past excitation divided by the square root of * the energy of filtered excitation. * corr[k] = <x[], y_k[]>/sqrt(y_k[],y_k[]) * where x[] is the target vector and y_k[] is the filtered past * excitation at delay k. * *************************************************************************/ static void Norm_Corr (Word16 exc[], Word16 xn[], Word16 h[], Word16 L_subfr, Word16 t_min, Word16 t_max, Word16 corr_norm[]) { Word16 i, j, k; Word16 corr_h, corr_l, norm_h, norm_l; Word32 s; /* Usally dynamic allocation of (L_subfr) */ Word16 excf[L_SUBFR]; Word16 scaling, h_fac, *s_excf, scaled_excf[L_SUBFR]; k = -t_min; move16 (); /* compute the filtered excitation for the first delay t_min */ Convolve (&exc[k], h, excf, L_subfr); /* scale "excf[]" to avoid overflow */ for (j = 0; j < L_subfr; j++) { scaled_excf[j] = shr (excf[j], 2); move16 (); } /* Compute 1/sqrt(energy of excf[]) */ s = 0; move32 (); for (j = 0; j < L_subfr; j++) { s = L_mac (s, excf[j], excf[j]); } test (); if (L_sub (s, 67108864L) <= 0) { /* if (s <= 2^26) */ s_excf = excf; move16 (); h_fac = 15 - 12; move16 (); scaling = 0; move16 (); } else { /* "excf[]" is divided by 2 */ s_excf = scaled_excf; move16 (); h_fac = 15 - 12 - 2; move16 (); scaling = 2; move16 (); } /* loop for every possible period */ for (i = t_min; i <= t_max; i++) { /* Compute 1/sqrt(energy of excf[]) */ s = 0; move32 (); for (j = 0; j < L_subfr; j++) { s = L_mac (s, s_excf[j], s_excf[j]); } s = Inv_sqrt (s); L_Extract (s, &norm_h, &norm_l); /* Compute correlation between xn[] and excf[] */ s = 0; move32 (); for (j = 0; j < L_subfr; j++) { s = L_mac (s, xn[j], s_excf[j]); } L_Extract (s, &corr_h, &corr_l); /* Normalize correlation = correlation * (1/sqrt(energy)) */ s = Mpy_32 (corr_h, corr_l, norm_h, norm_l); corr_norm[i] = extract_h (L_shl (s, 16)); move16 (); /* modify the filtered excitation excf[] for the next iteration */ test (); if (sub (i, t_max) != 0) { k--; for (j = L_subfr - 1; j > 0; j--) { s = L_mult (exc[k], h[j]); s = L_shl (s, h_fac); s_excf[j] = add (extract_h (s), s_excf[j - 1]); move16 (); } s_excf[0] = shr (exc[k], scaling); move16 (); } } return; } /************************************************************************* * * FUNCTION: searchFrac() * * PURPOSE: Find fractional pitch * * DESCRIPTION: * The function interpolates the normalized correlation at the * fractional positions around lag T0. The position at which the * interpolation function reaches its maximum is the fractional pitch. * Starting point of the search is frac, end point is last_frac. * frac is overwritten with the fractional pitch. * *************************************************************************/ static void searchFrac ( Word16 *lag, /* i/o : integer pitch */ Word16 *frac, /* i/o : start point of search - fractional pitch */ Word16 last_frac, /* i : endpoint of search */ Word16 corr[], /* i : normalized correlation */ Word16 flag3 /* i : subsample resolution (3: =1 / 6: =0) */ ) { Word16 i; Word16 max; Word16 corr_int; /* Test the fractions around T0 and choose the one which maximizes */ /* the interpolated normalized correlation. */ max = Interpol_3or6 (&corr[*lag], *frac, flag3); move16 (); /* function result */ for (i = add (*frac, 1); i <= last_frac; i++) { corr_int = Interpol_3or6 (&corr[*lag], i, flag3); move16 (); test (); if (sub (corr_int, max) > 0) { max = corr_int; move16 (); *frac = i; move16 (); } } test(); if (flag3 == 0) { /* Limit the fraction value in the interval [-2,-1,0,1,2,3] */ test (); if (sub (*frac, -3) == 0) { *frac = 3; move16 (); *lag = sub (*lag, 1); } } else { /* limit the fraction value between -1 and 1 */ test (); if (sub (*frac, -2) == 0) { *frac = 1; move16 (); *lag = sub (*lag, 1); } test (); if (sub (*frac, 2) == 0) { *frac = -1; move16 (); *lag = add (*lag, 1); } } } /************************************************************************* * * FUNCTION: getRange() * * PURPOSE: Sets range around open-loop pitch or integer pitch of last subframe * * DESCRIPTION: * Takes integer pitch T0 and calculates a range around it with * t0_min = T0-delta_low and t0_max = (T0-delta_low) + delta_range * t0_min and t0_max are bounded by pitmin and pitmax * *************************************************************************/ static void getRange ( Word16 T0, /* i : integer pitch */ Word16 delta_low, /* i : search start offset */ Word16 delta_range, /* i : search range */ Word16 pitmin, /* i : minimum pitch */ Word16 pitmax, /* i : maximum pitch */ Word16 *t0_min, /* o : search range minimum */ Word16 *t0_max) /* o : search range maximum */ { *t0_min = sub(T0, delta_low); test (); if (sub(*t0_min, pitmin) < 0) { *t0_min = pitmin; move16(); } *t0_max = add(*t0_min, delta_range); test (); if (sub(*t0_max, pitmax) > 0) { *t0_max = pitmax; move16(); *t0_min = sub(*t0_max, delta_range); } } /* ******************************************************************************** * PUBLIC PROGRAM CODE ******************************************************************************** */ /************************************************************************* * * Function: Pitch_fr_reset * Purpose: Initializes state memory to zero * ************************************************************************** */ void Pitch_fr_reset (Pitch_frState *state) { state->T0_prev_subframe = 0; } /************************************************************************* * * FUNCTION: Pitch_fr() * * PURPOSE: Find the pitch period with 1/3 or 1/6 subsample resolution * (closed loop). * * DESCRIPTION: * - find the normalized correlation between the target and filtered * past excitation in the search range. * - select the delay with maximum normalized correlation. * - interpolate the normalized correlation at fractions -3/6 to 3/6 * with step 1/6 around the chosen delay. * - The fraction which gives the maximum interpolated value is chosen. * *************************************************************************/ Word16 Pitch_fr ( /* o : pitch period (integer) */ Pitch_frState *st, /* i/o : State struct */ enum Mode mode, /* i : codec mode */ Word16 T_op[], /* i : open loop pitch lags */ Word16 exc[], /* i : excitation buffer Q0 */ Word16 xn[], /* i : target vector Q0 */ Word16 h[], /* i : impulse response of synthesis and weighting filters Q12 */ Word16 L_subfr, /* i : Length of subframe */ Word16 i_subfr, /* i : subframe offset */ Word16 *pit_frac, /* o : pitch period (fractional) */ Word16 *resu3, /* o : subsample resolution 1/3 (=1) or 1/6 (=0) */ Word16 *ana_index /* o : index of encoding */ ) { Word16 i; Word16 t_min, t_max; Word16 t0_min, t0_max; Word16 max, lag, frac; Word16 tmp_lag; Word16 *corr; Word16 corr_v[40]; /* Total length = t0_max-t0_min+1+2*L_INTER_SRCH */ Word16 max_frac_lag; Word16 flag3, flag4; Word16 last_frac; Word16 delta_int_low, delta_int_range; Word16 delta_frc_low, delta_frc_range; Word16 pit_min; Word16 frame_offset; Word16 delta_search; /*-----------------------------------------------------------------------* * set mode specific variables * *-----------------------------------------------------------------------*/ max_frac_lag = mode_dep_parm[mode].max_frac_lag; move16 (); flag3 = mode_dep_parm[mode].flag3; move16 (); frac = mode_dep_parm[mode].first_frac; move16 (); last_frac = mode_dep_parm[mode].last_frac; move16 (); delta_int_low = mode_dep_parm[mode].delta_int_low; move16 (); delta_int_range = mode_dep_parm[mode].delta_int_range; move16 (); delta_frc_low = mode_dep_parm[mode].delta_frc_low; move16 (); delta_frc_range = mode_dep_parm[mode].delta_frc_range; move16 (); pit_min = mode_dep_parm[mode].pit_min; move16 (); /*-----------------------------------------------------------------------* * decide upon full or differential search * *-----------------------------------------------------------------------*/ delta_search = 1; move16 (); test (); test (); if ((i_subfr == 0) || (sub(i_subfr,L_FRAME_BY2) == 0)) { /* Subframe 1 and 3 */ test (); test (); test (); if (((sub(mode, MR475) != 0) && (sub(mode, MR515) != 0)) || (sub(i_subfr,L_FRAME_BY2) != 0)) { /* set t0_min, t0_max for full search */ /* this is *not* done for mode MR475, MR515 in subframe 3 */ delta_search = 0; /* no differential search */ move16 (); /* calculate index into T_op which contains the open-loop */ /* pitch estimations for the 2 big subframes */ frame_offset = 1; move16 (); test (); if (i_subfr == 0) frame_offset = 0; move16 (); /* get T_op from the corresponding half frame and */ /* set t0_min, t0_max */ getRange (T_op[frame_offset], delta_int_low, delta_int_range, pit_min, PIT_MAX, &t0_min, &t0_max); } else { /* mode MR475, MR515 and 3. Subframe: delta search as well */ getRange (st->T0_prev_subframe, delta_frc_low, delta_frc_range, pit_min, PIT_MAX, &t0_min, &t0_max); } } else { /* for Subframe 2 and 4 */ /* get range around T0 of previous subframe for delta search */ getRange (st->T0_prev_subframe, delta_frc_low, delta_frc_range, pit_min, PIT_MAX, &t0_min, &t0_max); } /*-----------------------------------------------------------------------* * Find interval to compute normalized correlation * *-----------------------------------------------------------------------*/ t_min = sub (t0_min, L_INTER_SRCH); t_max = add (t0_max, L_INTER_SRCH); corr = &corr_v[-t_min]; move16 (); /*-----------------------------------------------------------------------* * Compute normalized correlation between target and filtered excitation * *-----------------------------------------------------------------------*/ Norm_Corr (exc, xn, h, L_subfr, t_min, t_max, corr); /*-----------------------------------------------------------------------* * Find integer pitch * *-----------------------------------------------------------------------*/ max = corr[t0_min]; move16 (); lag = t0_min; move16 (); for (i = t0_min + 1; i <= t0_max; i++) { test (); if (sub (corr[i], max) >= 0) { max = corr[i]; move16 (); lag = i; move16 (); } } /*-----------------------------------------------------------------------* * Find fractional pitch * *-----------------------------------------------------------------------*/ test (); test (); if ((delta_search == 0) && (sub (lag, max_frac_lag) > 0)) { /* full search and integer pitch greater than max_frac_lag */ /* fractional search is not needed, set fractional to zero */ frac = 0; move16 (); } else { /* if differential search AND mode MR475 OR MR515 OR MR59 OR MR67 */ /* then search fractional with 4 bits resolution */ test (); test (); test (); test (); test (); if ((delta_search != 0) && ((sub (mode, MR475) == 0) || (sub (mode, MR515) == 0) || (sub (mode, MR59) == 0) || (sub (mode, MR67) == 0))) { /* modify frac or last_frac according to position of last */ /* integer pitch: either search around integer pitch, */ /* or only on left or right side */ tmp_lag = st->T0_prev_subframe; move16 (); test (); if ( sub( sub(tmp_lag, t0_min), 5) > 0) tmp_lag = add (t0_min, 5); test (); if ( sub( sub(t0_max, tmp_lag), 4) > 0) tmp_lag = sub (t0_max, 4); test (); test (); if ((sub (lag, tmp_lag) == 0) || (sub (lag, sub(tmp_lag, 1)) == 0)) { /* normal search in fractions around T0 */ searchFrac (&lag, &frac, last_frac, corr, flag3); } else if (sub (lag, sub (tmp_lag, 2)) == 0) { test (); /* limit search around T0 to the right side */ frac = 0; move16 (); searchFrac (&lag, &frac, last_frac, corr, flag3); } else if (sub (lag, add(tmp_lag, 1)) == 0) { test (); test (); /* limit search around T0 to the left side */ last_frac = 0; move16 (); searchFrac (&lag, &frac, last_frac, corr, flag3); } else { test (); test (); /* no fractional search */ frac = 0; move16 (); } } else /* test the fractions around T0 */ searchFrac (&lag, &frac, last_frac, corr, flag3); } /*-----------------------------------------------------------------------* * encode pitch * *-----------------------------------------------------------------------*/ test (); if (flag3 != 0) { /* flag4 indicates encoding with 4 bit resolution; */ /* this is needed for mode MR475, MR515 and MR59 */ flag4 = 0; move16 (); test (); test (); test (); test (); if ( (sub (mode, MR475) == 0) || (sub (mode, MR515) == 0) || (sub (mode, MR59) == 0) || (sub (mode, MR67) == 0) ) { flag4 = 1; move16 (); } /* encode with 1/3 subsample resolution */ *ana_index = Enc_lag3(lag, frac, st->T0_prev_subframe, t0_min, t0_max, delta_search, flag4); move16 (); /* function result */ } else { /* encode with 1/6 subsample resolution */ *ana_index = Enc_lag6(lag, frac, t0_min, delta_search); move16 (); /* function result */ } /*-----------------------------------------------------------------------* * update state variables * *-----------------------------------------------------------------------*/ st->T0_prev_subframe = lag; move16 (); /*-----------------------------------------------------------------------* * update output variables * *-----------------------------------------------------------------------*/ *resu3 = flag3; move16 (); *pit_frac = frac; move16 (); return (lag); }