Mercurial > hg > audiostuff
view spandsp-0.0.6pre17/src/g726.c @ 4:26cd8f1ef0b1
import spandsp-0.0.6pre17
author | Peter Meerwald <pmeerw@cosy.sbg.ac.at> |
---|---|
date | Fri, 25 Jun 2010 15:50:58 +0200 |
parents | |
children |
line wrap: on
line source
/* * SpanDSP - a series of DSP components for telephony * * g726.c - The ITU G.726 codec. * * Written by Steve Underwood <steveu@coppice.org> * * Copyright (C) 2006 Steve Underwood * * All rights reserved. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU Lesser General Public License version 2.1, * as published by the Free Software Foundation. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with this program; if not, write to the Free Software * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. * * Based on G.721/G.723 code which is: * * This source code is a product of Sun Microsystems, Inc. and is provided * for unrestricted use. Users may copy or modify this source code without * charge. * * SUN SOURCE CODE IS PROVIDED AS IS WITH NO WARRANTIES OF ANY KIND INCLUDING * THE WARRANTIES OF DESIGN, MERCHANTIBILITY AND FITNESS FOR A PARTICULAR * PURPOSE, OR ARISING FROM A COURSE OF DEALING, USAGE OR TRADE PRACTICE. * * Sun source code is provided with no support and without any obligation on * the part of Sun Microsystems, Inc. to assist in its use, correction, * modification or enhancement. * * SUN MICROSYSTEMS, INC. SHALL HAVE NO LIABILITY WITH RESPECT TO THE * INFRINGEMENT OF COPYRIGHTS, TRADE SECRETS OR ANY PATENTS BY THIS SOFTWARE * OR ANY PART THEREOF. * * In no event will Sun Microsystems, Inc. be liable for any lost revenue * or profits or other special, indirect and consequential damages, even if * Sun has been advised of the possibility of such damages. * * Sun Microsystems, Inc. * 2550 Garcia Avenue * Mountain View, California 94043 * * $Id: g726.c,v 1.28.4.1 2009/12/28 12:20:46 steveu Exp $ */ /*! \file */ #if defined(HAVE_CONFIG_H) #include "config.h" #endif #include <inttypes.h> #include <memory.h> #include <stdlib.h> #if defined(HAVE_TGMATH_H) #include <tgmath.h> #endif #if defined(HAVE_MATH_H) #include <math.h> #endif #include "floating_fudge.h" #include "spandsp/telephony.h" #include "spandsp/dc_restore.h" #include "spandsp/bitstream.h" #include "spandsp/bit_operations.h" #include "spandsp/g711.h" #include "spandsp/g726.h" #include "spandsp/private/bitstream.h" #include "spandsp/private/g726.h" /* * Maps G.726_16 code word to reconstructed scale factor normalized log * magnitude values. */ static const int g726_16_dqlntab[4] = { 116, 365, 365, 116 }; /* Maps G.726_16 code word to log of scale factor multiplier. */ static const int g726_16_witab[4] = { -704, 14048, 14048, -704 }; /* * Maps G.726_16 code words to a set of values whose long and short * term averages are computed and then compared to give an indication * how stationary (steady state) the signal is. */ static const int g726_16_fitab[4] = { 0x000, 0xE00, 0xE00, 0x000 }; static const int qtab_726_16[1] = { 261 }; /* * Maps G.726_24 code word to reconstructed scale factor normalized log * magnitude values. */ static const int g726_24_dqlntab[8] = { -2048, 135, 273, 373, 373, 273, 135, -2048 }; /* Maps G.726_24 code word to log of scale factor multiplier. */ static const int g726_24_witab[8] = { -128, 960, 4384, 18624, 18624, 4384, 960, -128 }; /* * Maps G.726_24 code words to a set of values whose long and short * term averages are computed and then compared to give an indication * how stationary (steady state) the signal is. */ static const int g726_24_fitab[8] = { 0x000, 0x200, 0x400, 0xE00, 0xE00, 0x400, 0x200, 0x000 }; static const int qtab_726_24[3] = { 8, 218, 331 }; /* * Maps G.726_32 code word to reconstructed scale factor normalized log * magnitude values. */ static const int g726_32_dqlntab[16] = { -2048, 4, 135, 213, 273, 323, 373, 425, 425, 373, 323, 273, 213, 135, 4, -2048 }; /* Maps G.726_32 code word to log of scale factor multiplier. */ static const int g726_32_witab[16] = { -384, 576, 1312, 2048, 3584, 6336, 11360, 35904, 35904, 11360, 6336, 3584, 2048, 1312, 576, -384 }; /* * Maps G.726_32 code words to a set of values whose long and short * term averages are computed and then compared to give an indication * how stationary (steady state) the signal is. */ static const int g726_32_fitab[16] = { 0x000, 0x000, 0x000, 0x200, 0x200, 0x200, 0x600, 0xE00, 0xE00, 0x600, 0x200, 0x200, 0x200, 0x000, 0x000, 0x000 }; static const int qtab_726_32[7] = { -124, 80, 178, 246, 300, 349, 400 }; /* * Maps G.726_40 code word to ructeconstructed scale factor normalized log * magnitude values. */ static const int g726_40_dqlntab[32] = { -2048, -66, 28, 104, 169, 224, 274, 318, 358, 395, 429, 459, 488, 514, 539, 566, 566, 539, 514, 488, 459, 429, 395, 358, 318, 274, 224, 169, 104, 28, -66, -2048 }; /* Maps G.726_40 code word to log of scale factor multiplier. */ static const int g726_40_witab[32] = { 448, 448, 768, 1248, 1280, 1312, 1856, 3200, 4512, 5728, 7008, 8960, 11456, 14080, 16928, 22272, 22272, 16928, 14080, 11456, 8960, 7008, 5728, 4512, 3200, 1856, 1312, 1280, 1248, 768, 448, 448 }; /* * Maps G.726_40 code words to a set of values whose long and short * term averages are computed and then compared to give an indication * how stationary (steady state) the signal is. */ static const int g726_40_fitab[32] = { 0x000, 0x000, 0x000, 0x000, 0x000, 0x200, 0x200, 0x200, 0x200, 0x200, 0x400, 0x600, 0x800, 0xA00, 0xC00, 0xC00, 0xC00, 0xC00, 0xA00, 0x800, 0x600, 0x400, 0x200, 0x200, 0x200, 0x200, 0x200, 0x000, 0x000, 0x000, 0x000, 0x000 }; static const int qtab_726_40[15] = { -122, -16, 68, 139, 198, 250, 298, 339, 378, 413, 445, 475, 502, 528, 553 }; /* * returns the integer product of the 14-bit integer "an" and * "floating point" representation (4-bit exponent, 6-bit mantessa) "srn". */ static int16_t fmult(int16_t an, int16_t srn) { int16_t anmag; int16_t anexp; int16_t anmant; int16_t wanexp; int16_t wanmant; int16_t retval; anmag = (an > 0) ? an : ((-an) & 0x1FFF); anexp = (int16_t) (top_bit(anmag) - 5); anmant = (anmag == 0) ? 32 : (anexp >= 0) ? (anmag >> anexp) : (anmag << -anexp); wanexp = anexp + ((srn >> 6) & 0xF) - 13; wanmant = (anmant*(srn & 0x3F) + 0x30) >> 4; retval = (wanexp >= 0) ? ((wanmant << wanexp) & 0x7FFF) : (wanmant >> -wanexp); return (((an ^ srn) < 0) ? -retval : retval); } /*- End of function --------------------------------------------------------*/ /* * Compute the estimated signal from the 6-zero predictor. */ static __inline__ int16_t predictor_zero(g726_state_t *s) { int i; int sezi; sezi = fmult(s->b[0] >> 2, s->dq[0]); /* ACCUM */ for (i = 1; i < 6; i++) sezi += fmult(s->b[i] >> 2, s->dq[i]); return (int16_t) sezi; } /*- End of function --------------------------------------------------------*/ /* * Computes the estimated signal from the 2-pole predictor. */ static __inline__ int16_t predictor_pole(g726_state_t *s) { return (fmult(s->a[1] >> 2, s->sr[1]) + fmult(s->a[0] >> 2, s->sr[0])); } /*- End of function --------------------------------------------------------*/ /* * Computes the quantization step size of the adaptive quantizer. */ static int step_size(g726_state_t *s) { int y; int dif; int al; if (s->ap >= 256) return s->yu; y = s->yl >> 6; dif = s->yu - y; al = s->ap >> 2; if (dif > 0) y += (dif*al) >> 6; else if (dif < 0) y += (dif*al + 0x3F) >> 6; return y; } /*- End of function --------------------------------------------------------*/ /* * Given a raw sample, 'd', of the difference signal and a * quantization step size scale factor, 'y', this routine returns the * ADPCM codeword to which that sample gets quantized. The step * size scale factor division operation is done in the log base 2 domain * as a subtraction. */ static int16_t quantize(int d, /* Raw difference signal sample */ int y, /* Step size multiplier */ const int table[], /* quantization table */ int quantizer_states) /* table size of int16_t integers */ { int16_t dqm; /* Magnitude of 'd' */ int16_t exp; /* Integer part of base 2 log of 'd' */ int16_t mant; /* Fractional part of base 2 log */ int16_t dl; /* Log of magnitude of 'd' */ int16_t dln; /* Step size scale factor normalized log */ int i; int size; /* * LOG * * Compute base 2 log of 'd', and store in 'dl'. */ dqm = (int16_t) abs(d); exp = (int16_t) (top_bit(dqm >> 1) + 1); /* Fractional portion. */ mant = ((dqm << 7) >> exp) & 0x7F; dl = (exp << 7) + mant; /* * SUBTB * * "Divide" by step size multiplier. */ dln = dl - (int16_t) (y >> 2); /* * QUAN * * Search for codword i for 'dln'. */ size = (quantizer_states - 1) >> 1; for (i = 0; i < size; i++) { if (dln < table[i]) break; } if (d < 0) { /* Take 1's complement of i */ return (int16_t) ((size << 1) + 1 - i); } if (i == 0 && (quantizer_states & 1)) { /* Zero is only valid if there are an even number of states, so take the 1's complement if the code is zero. */ return (int16_t) quantizer_states; } return (int16_t) i; } /*- End of function --------------------------------------------------------*/ /* * Returns reconstructed difference signal 'dq' obtained from * codeword 'i' and quantization step size scale factor 'y'. * Multiplication is performed in log base 2 domain as addition. */ static int16_t reconstruct(int sign, /* 0 for non-negative value */ int dqln, /* G.72x codeword */ int y) /* Step size multiplier */ { int16_t dql; /* Log of 'dq' magnitude */ int16_t dex; /* Integer part of log */ int16_t dqt; int16_t dq; /* Reconstructed difference signal sample */ dql = (int16_t) (dqln + (y >> 2)); /* ADDA */ if (dql < 0) return ((sign) ? -0x8000 : 0); /* ANTILOG */ dex = (dql >> 7) & 15; dqt = 128 + (dql & 127); dq = (dqt << 7) >> (14 - dex); return ((sign) ? (dq - 0x8000) : dq); } /*- End of function --------------------------------------------------------*/ /* * updates the state variables for each output code */ static void update(g726_state_t *s, int y, /* quantizer step size */ int wi, /* scale factor multiplier */ int fi, /* for long/short term energies */ int dq, /* quantized prediction difference */ int sr, /* reconstructed signal */ int dqsez) /* difference from 2-pole predictor */ { int16_t mag; int16_t exp; int16_t a2p; /* LIMC */ int16_t a1ul; /* UPA1 */ int16_t pks1; /* UPA2 */ int16_t fa1; int16_t ylint; int16_t dqthr; int16_t ylfrac; int16_t thr; int16_t pk0; int i; int tr; a2p = 0; /* Needed in updating predictor poles */ pk0 = (dqsez < 0) ? 1 : 0; /* prediction difference magnitude */ mag = (int16_t) (dq & 0x7FFF); /* TRANS */ ylint = (int16_t) (s->yl >> 15); /* exponent part of yl */ ylfrac = (int16_t) ((s->yl >> 10) & 0x1F); /* fractional part of yl */ /* Limit threshold to 31 << 10 */ thr = (ylint > 9) ? (31 << 10) : ((32 + ylfrac) << ylint); dqthr = (thr + (thr >> 1)) >> 1; /* dqthr = 0.75 * thr */ if (!s->td) /* signal supposed voice */ tr = FALSE; else if (mag <= dqthr) /* supposed data, but small mag */ tr = FALSE; /* treated as voice */ else /* signal is data (modem) */ tr = TRUE; /* * Quantizer scale factor adaptation. */ /* FUNCTW & FILTD & DELAY */ /* update non-steady state step size multiplier */ s->yu = (int16_t) (y + ((wi - y) >> 5)); /* LIMB */ if (s->yu < 544) s->yu = 544; else if (s->yu > 5120) s->yu = 5120; /* FILTE & DELAY */ /* update steady state step size multiplier */ s->yl += s->yu + ((-s->yl) >> 6); /* * Adaptive predictor coefficients. */ if (tr) { /* Reset the a's and b's for a modem signal */ s->a[0] = 0; s->a[1] = 0; s->b[0] = 0; s->b[1] = 0; s->b[2] = 0; s->b[3] = 0; s->b[4] = 0; s->b[5] = 0; } else { /* Update the a's and b's */ /* UPA2 */ pks1 = pk0 ^ s->pk[0]; /* Update predictor pole a[1] */ a2p = s->a[1] - (s->a[1] >> 7); if (dqsez != 0) { fa1 = (pks1) ? s->a[0] : -s->a[0]; /* a2p = function of fa1 */ if (fa1 < -8191) a2p -= 0x100; else if (fa1 > 8191) a2p += 0xFF; else a2p += fa1 >> 5; if (pk0 ^ s->pk[1]) { /* LIMC */ if (a2p <= -12160) a2p = -12288; else if (a2p >= 12416) a2p = 12288; else a2p -= 0x80; } else if (a2p <= -12416) a2p = -12288; else if (a2p >= 12160) a2p = 12288; else a2p += 0x80; } /* TRIGB & DELAY */ s->a[1] = a2p; /* UPA1 */ /* Update predictor pole a[0] */ s->a[0] -= s->a[0] >> 8; if (dqsez != 0) { if (pks1 == 0) s->a[0] += 192; else s->a[0] -= 192; } /* LIMD */ a1ul = 15360 - a2p; if (s->a[0] < -a1ul) s->a[0] = -a1ul; else if (s->a[0] > a1ul) s->a[0] = a1ul; /* UPB : update predictor zeros b[6] */ for (i = 0; i < 6; i++) { /* Distinguish 40Kbps mode from the others */ s->b[i] -= s->b[i] >> ((s->bits_per_sample == 5) ? 9 : 8); if (dq & 0x7FFF) { /* XOR */ if ((dq ^ s->dq[i]) >= 0) s->b[i] += 128; else s->b[i] -= 128; } } } for (i = 5; i > 0; i--) s->dq[i] = s->dq[i - 1]; /* FLOAT A : convert dq[0] to 4-bit exp, 6-bit mantissa f.p. */ if (mag == 0) { s->dq[0] = (dq >= 0) ? 0x20 : 0xFC20; } else { exp = (int16_t) (top_bit(mag) + 1); s->dq[0] = (dq >= 0) ? ((exp << 6) + ((mag << 6) >> exp)) : ((exp << 6) + ((mag << 6) >> exp) - 0x400); } s->sr[1] = s->sr[0]; /* FLOAT B : convert sr to 4-bit exp., 6-bit mantissa f.p. */ if (sr == 0) { s->sr[0] = 0x20; } else if (sr > 0) { exp = (int16_t) (top_bit(sr) + 1); s->sr[0] = (int16_t) ((exp << 6) + ((sr << 6) >> exp)); } else if (sr > -32768) { mag = (int16_t) -sr; exp = (int16_t) (top_bit(mag) + 1); s->sr[0] = (exp << 6) + ((mag << 6) >> exp) - 0x400; } else { s->sr[0] = (uint16_t) 0xFC20; } /* DELAY A */ s->pk[1] = s->pk[0]; s->pk[0] = pk0; /* TONE */ if (tr) /* this sample has been treated as data */ s->td = FALSE; /* next one will be treated as voice */ else if (a2p < -11776) /* small sample-to-sample correlation */ s->td = TRUE; /* signal may be data */ else /* signal is voice */ s->td = FALSE; /* Adaptation speed control. */ /* FILTA */ s->dms += ((int16_t) fi - s->dms) >> 5; /* FILTB */ s->dml += (((int16_t) (fi << 2) - s->dml) >> 7); if (tr) s->ap = 256; else if (y < 1536) /* SUBTC */ s->ap += (0x200 - s->ap) >> 4; else if (s->td) s->ap += (0x200 - s->ap) >> 4; else if (abs((s->dms << 2) - s->dml) >= (s->dml >> 3)) s->ap += (0x200 - s->ap) >> 4; else s->ap += (-s->ap) >> 4; } /*- End of function --------------------------------------------------------*/ static int16_t tandem_adjust_alaw(int16_t sr, /* decoder output linear PCM sample */ int se, /* predictor estimate sample */ int y, /* quantizer step size */ int i, /* decoder input code */ int sign, const int qtab[], int quantizer_states) { uint8_t sp; /* A-law compressed 8-bit code */ int16_t dx; /* prediction error */ int id; /* quantized prediction error */ int sd; /* adjusted A-law decoded sample value */ if (sr <= -32768) sr = -1; sp = linear_to_alaw((sr >> 1) << 3); /* 16-bit prediction error */ dx = (int16_t) ((alaw_to_linear(sp) >> 2) - se); id = quantize(dx, y, qtab, quantizer_states); if (id == i) { /* No adjustment of sp required */ return (int16_t) sp; } /* sp adjustment needed */ /* ADPCM codes : 8, 9, ... F, 0, 1, ... , 6, 7 */ /* 2's complement to biased unsigned */ if ((id ^ sign) > (i ^ sign)) { /* sp adjusted to next lower value */ if (sp & 0x80) sd = (sp == 0xD5) ? 0x55 : (((sp ^ 0x55) - 1) ^ 0x55); else sd = (sp == 0x2A) ? 0x2A : (((sp ^ 0x55) + 1) ^ 0x55); } else { /* sp adjusted to next higher value */ if (sp & 0x80) sd = (sp == 0xAA) ? 0xAA : (((sp ^ 0x55) + 1) ^ 0x55); else sd = (sp == 0x55) ? 0xD5 : (((sp ^ 0x55) - 1) ^ 0x55); } return (int16_t) sd; } /*- End of function --------------------------------------------------------*/ static int16_t tandem_adjust_ulaw(int16_t sr, /* decoder output linear PCM sample */ int se, /* predictor estimate sample */ int y, /* quantizer step size */ int i, /* decoder input code */ int sign, const int qtab[], int quantizer_states) { uint8_t sp; /* u-law compressed 8-bit code */ int16_t dx; /* prediction error */ int id; /* quantized prediction error */ int sd; /* adjusted u-law decoded sample value */ if (sr <= -32768) sr = 0; sp = linear_to_ulaw(sr << 2); /* 16-bit prediction error */ dx = (int16_t) ((ulaw_to_linear(sp) >> 2) - se); id = quantize(dx, y, qtab, quantizer_states); if (id == i) { /* No adjustment of sp required. */ return (int16_t) sp; } /* ADPCM codes : 8, 9, ... F, 0, 1, ... , 6, 7 */ /* 2's complement to biased unsigned */ if ((id ^ sign) > (i ^ sign)) { /* sp adjusted to next lower value */ if (sp & 0x80) sd = (sp == 0xFF) ? 0x7E : (sp + 1); else sd = (sp == 0x00) ? 0x00 : (sp - 1); } else { /* sp adjusted to next higher value */ if (sp & 0x80) sd = (sp == 0x80) ? 0x80 : (sp - 1); else sd = (sp == 0x7F) ? 0xFE : (sp + 1); } return (int16_t) sd; } /*- End of function --------------------------------------------------------*/ /* * Encodes a linear PCM, A-law or u-law input sample and returns its 3-bit code. */ static uint8_t g726_16_encoder(g726_state_t *s, int16_t amp) { int y; int16_t sei; int16_t sezi; int16_t se; int16_t d; int16_t sr; int16_t dqsez; int16_t dq; int16_t i; sezi = predictor_zero(s); sei = sezi + predictor_pole(s); se = sei >> 1; d = amp - se; /* Quantize prediction difference */ y = step_size(s); i = quantize(d, y, qtab_726_16, 4); dq = reconstruct(i & 2, g726_16_dqlntab[i], y); /* Reconstruct the signal */ sr = (dq < 0) ? (se - (dq & 0x3FFF)) : (se + dq); /* Pole prediction difference */ dqsez = sr + (sezi >> 1) - se; update(s, y, g726_16_witab[i], g726_16_fitab[i], dq, sr, dqsez); return (uint8_t) i; } /*- End of function --------------------------------------------------------*/ /* * Decodes a 2-bit CCITT G.726_16 ADPCM code and returns * the resulting 16-bit linear PCM, A-law or u-law sample value. */ static int16_t g726_16_decoder(g726_state_t *s, uint8_t code) { int16_t sezi; int16_t sei; int16_t se; int16_t sr; int16_t dq; int16_t dqsez; int y; /* Mask to get proper bits */ code &= 0x03; sezi = predictor_zero(s); sei = sezi + predictor_pole(s); y = step_size(s); dq = reconstruct(code & 2, g726_16_dqlntab[code], y); /* Reconstruct the signal */ se = sei >> 1; sr = (dq < 0) ? (se - (dq & 0x3FFF)) : (se + dq); /* Pole prediction difference */ dqsez = sr + (sezi >> 1) - se; update(s, y, g726_16_witab[code], g726_16_fitab[code], dq, sr, dqsez); switch (s->ext_coding) { case G726_ENCODING_ALAW: return tandem_adjust_alaw(sr, se, y, code, 2, qtab_726_16, 4); case G726_ENCODING_ULAW: return tandem_adjust_ulaw(sr, se, y, code, 2, qtab_726_16, 4); } return (sr << 2); } /*- End of function --------------------------------------------------------*/ /* * Encodes a linear PCM, A-law or u-law input sample and returns its 3-bit code. */ static uint8_t g726_24_encoder(g726_state_t *s, int16_t amp) { int16_t sei; int16_t sezi; int16_t se; int16_t d; int16_t sr; int16_t dqsez; int16_t dq; int16_t i; int y; sezi = predictor_zero(s); sei = sezi + predictor_pole(s); se = sei >> 1; d = amp - se; /* Quantize prediction difference */ y = step_size(s); i = quantize(d, y, qtab_726_24, 7); dq = reconstruct(i & 4, g726_24_dqlntab[i], y); /* Reconstruct the signal */ sr = (dq < 0) ? (se - (dq & 0x3FFF)) : (se + dq); /* Pole prediction difference */ dqsez = sr + (sezi >> 1) - se; update(s, y, g726_24_witab[i], g726_24_fitab[i], dq, sr, dqsez); return (uint8_t) i; } /*- End of function --------------------------------------------------------*/ /* * Decodes a 3-bit CCITT G.726_24 ADPCM code and returns * the resulting 16-bit linear PCM, A-law or u-law sample value. */ static int16_t g726_24_decoder(g726_state_t *s, uint8_t code) { int16_t sezi; int16_t sei; int16_t se; int16_t sr; int16_t dq; int16_t dqsez; int y; /* Mask to get proper bits */ code &= 0x07; sezi = predictor_zero(s); sei = sezi + predictor_pole(s); y = step_size(s); dq = reconstruct(code & 4, g726_24_dqlntab[code], y); /* Reconstruct the signal */ se = sei >> 1; sr = (dq < 0) ? (se - (dq & 0x3FFF)) : (se + dq); /* Pole prediction difference */ dqsez = sr + (sezi >> 1) - se; update(s, y, g726_24_witab[code], g726_24_fitab[code], dq, sr, dqsez); switch (s->ext_coding) { case G726_ENCODING_ALAW: return tandem_adjust_alaw(sr, se, y, code, 4, qtab_726_24, 7); case G726_ENCODING_ULAW: return tandem_adjust_ulaw(sr, se, y, code, 4, qtab_726_24, 7); } return (sr << 2); } /*- End of function --------------------------------------------------------*/ /* * Encodes a linear input sample and returns its 4-bit code. */ static uint8_t g726_32_encoder(g726_state_t *s, int16_t amp) { int16_t sei; int16_t sezi; int16_t se; int16_t d; int16_t sr; int16_t dqsez; int16_t dq; int16_t i; int y; sezi = predictor_zero(s); sei = sezi + predictor_pole(s); se = sei >> 1; d = amp - se; /* Quantize the prediction difference */ y = step_size(s); i = quantize(d, y, qtab_726_32, 15); dq = reconstruct(i & 8, g726_32_dqlntab[i], y); /* Reconstruct the signal */ sr = (dq < 0) ? (se - (dq & 0x3FFF)) : (se + dq); /* Pole prediction difference */ dqsez = sr + (sezi >> 1) - se; update(s, y, g726_32_witab[i], g726_32_fitab[i], dq, sr, dqsez); return (uint8_t) i; } /*- End of function --------------------------------------------------------*/ /* * Decodes a 4-bit CCITT G.726_32 ADPCM code and returns * the resulting 16-bit linear PCM, A-law or u-law sample value. */ static int16_t g726_32_decoder(g726_state_t *s, uint8_t code) { int16_t sezi; int16_t sei; int16_t se; int16_t sr; int16_t dq; int16_t dqsez; int y; /* Mask to get proper bits */ code &= 0x0F; sezi = predictor_zero(s); sei = sezi + predictor_pole(s); y = step_size(s); dq = reconstruct(code & 8, g726_32_dqlntab[code], y); /* Reconstruct the signal */ se = sei >> 1; sr = (dq < 0) ? (se - (dq & 0x3FFF)) : (se + dq); /* Pole prediction difference */ dqsez = sr + (sezi >> 1) - se; update(s, y, g726_32_witab[code], g726_32_fitab[code], dq, sr, dqsez); switch (s->ext_coding) { case G726_ENCODING_ALAW: return tandem_adjust_alaw(sr, se, y, code, 8, qtab_726_32, 15); case G726_ENCODING_ULAW: return tandem_adjust_ulaw(sr, se, y, code, 8, qtab_726_32, 15); } return (sr << 2); } /*- End of function --------------------------------------------------------*/ /* * Encodes a 16-bit linear PCM, A-law or u-law input sample and retuens * the resulting 5-bit CCITT G.726 40Kbps code. */ static uint8_t g726_40_encoder(g726_state_t *s, int16_t amp) { int16_t sei; int16_t sezi; int16_t se; int16_t d; int16_t sr; int16_t dqsez; int16_t dq; int16_t i; int y; sezi = predictor_zero(s); sei = sezi + predictor_pole(s); se = sei >> 1; d = amp - se; /* Quantize prediction difference */ y = step_size(s); i = quantize(d, y, qtab_726_40, 31); dq = reconstruct(i & 0x10, g726_40_dqlntab[i], y); /* Reconstruct the signal */ sr = (dq < 0) ? (se - (dq & 0x7FFF)) : (se + dq); /* Pole prediction difference */ dqsez = sr + (sezi >> 1) - se; update(s, y, g726_40_witab[i], g726_40_fitab[i], dq, sr, dqsez); return (uint8_t) i; } /*- End of function --------------------------------------------------------*/ /* * Decodes a 5-bit CCITT G.726 40Kbps code and returns * the resulting 16-bit linear PCM, A-law or u-law sample value. */ static int16_t g726_40_decoder(g726_state_t *s, uint8_t code) { int16_t sezi; int16_t sei; int16_t se; int16_t sr; int16_t dq; int16_t dqsez; int y; /* Mask to get proper bits */ code &= 0x1F; sezi = predictor_zero(s); sei = sezi + predictor_pole(s); y = step_size(s); dq = reconstruct(code & 0x10, g726_40_dqlntab[code], y); /* Reconstruct the signal */ se = sei >> 1; sr = (dq < 0) ? (se - (dq & 0x7FFF)) : (se + dq); /* Pole prediction difference */ dqsez = sr + (sezi >> 1) - se; update(s, y, g726_40_witab[code], g726_40_fitab[code], dq, sr, dqsez); switch (s->ext_coding) { case G726_ENCODING_ALAW: return tandem_adjust_alaw(sr, se, y, code, 0x10, qtab_726_40, 31); case G726_ENCODING_ULAW: return tandem_adjust_ulaw(sr, se, y, code, 0x10, qtab_726_40, 31); } return (sr << 2); } /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(g726_state_t *) g726_init(g726_state_t *s, int bit_rate, int ext_coding, int packing) { int i; if (bit_rate != 16000 && bit_rate != 24000 && bit_rate != 32000 && bit_rate != 40000) return NULL; if (s == NULL) { if ((s = (g726_state_t *) malloc(sizeof(*s))) == NULL) return NULL; } s->yl = 34816; s->yu = 544; s->dms = 0; s->dml = 0; s->ap = 0; s->rate = bit_rate; s->ext_coding = ext_coding; s->packing = packing; for (i = 0; i < 2; i++) { s->a[i] = 0; s->pk[i] = 0; s->sr[i] = 32; } for (i = 0; i < 6; i++) { s->b[i] = 0; s->dq[i] = 32; } s->td = FALSE; switch (bit_rate) { case 16000: s->enc_func = g726_16_encoder; s->dec_func = g726_16_decoder; s->bits_per_sample = 2; break; case 24000: s->enc_func = g726_24_encoder; s->dec_func = g726_24_decoder; s->bits_per_sample = 3; break; case 32000: default: s->enc_func = g726_32_encoder; s->dec_func = g726_32_decoder; s->bits_per_sample = 4; break; case 40000: s->enc_func = g726_40_encoder; s->dec_func = g726_40_decoder; s->bits_per_sample = 5; break; } bitstream_init(&s->bs, (s->packing != G726_PACKING_LEFT)); return s; } /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(int) g726_release(g726_state_t *s) { return 0; } /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(int) g726_free(g726_state_t *s) { free(s); return 0; } /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(int) g726_decode(g726_state_t *s, int16_t amp[], const uint8_t g726_data[], int g726_bytes) { int i; int samples; uint8_t code; int sl; for (samples = i = 0; ; ) { if (s->packing != G726_PACKING_NONE) { /* Unpack the code bits */ if (s->packing != G726_PACKING_LEFT) { if (s->bs.residue < s->bits_per_sample) { if (i >= g726_bytes) break; s->bs.bitstream |= (g726_data[i++] << s->bs.residue); s->bs.residue += 8; } code = (uint8_t) (s->bs.bitstream & ((1 << s->bits_per_sample) - 1)); s->bs.bitstream >>= s->bits_per_sample; } else { if (s->bs.residue < s->bits_per_sample) { if (i >= g726_bytes) break; s->bs.bitstream = (s->bs.bitstream << 8) | g726_data[i++]; s->bs.residue += 8; } code = (uint8_t) ((s->bs.bitstream >> (s->bs.residue - s->bits_per_sample)) & ((1 << s->bits_per_sample) - 1)); } s->bs.residue -= s->bits_per_sample; } else { if (i >= g726_bytes) break; code = g726_data[i++]; } sl = s->dec_func(s, code); if (s->ext_coding != G726_ENCODING_LINEAR) ((uint8_t *) amp)[samples++] = (uint8_t) sl; else amp[samples++] = (int16_t) sl; } return samples; } /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(int) g726_encode(g726_state_t *s, uint8_t g726_data[], const int16_t amp[], int len) { int i; int g726_bytes; int16_t sl; uint8_t code; for (g726_bytes = i = 0; i < len; i++) { /* Linearize the input sample to 14-bit PCM */ switch (s->ext_coding) { case G726_ENCODING_ALAW: sl = alaw_to_linear(((const uint8_t *) amp)[i]) >> 2; break; case G726_ENCODING_ULAW: sl = ulaw_to_linear(((const uint8_t *) amp)[i]) >> 2; break; default: sl = amp[i] >> 2; break; } code = s->enc_func(s, sl); if (s->packing != G726_PACKING_NONE) { /* Pack the code bits */ if (s->packing != G726_PACKING_LEFT) { s->bs.bitstream |= (code << s->bs.residue); s->bs.residue += s->bits_per_sample; if (s->bs.residue >= 8) { g726_data[g726_bytes++] = (uint8_t) (s->bs.bitstream & 0xFF); s->bs.bitstream >>= 8; s->bs.residue -= 8; } } else { s->bs.bitstream = (s->bs.bitstream << s->bits_per_sample) | code; s->bs.residue += s->bits_per_sample; if (s->bs.residue >= 8) { g726_data[g726_bytes++] = (uint8_t) ((s->bs.bitstream >> (s->bs.residue - 8)) & 0xFF); s->bs.residue -= 8; } } } else { g726_data[g726_bytes++] = (uint8_t) code; } } return g726_bytes; } /*- End of function --------------------------------------------------------*/ /*- End of file ------------------------------------------------------------*/