Add G.726-32kbps Codec Transcoder (Tested with Cisco ATA-186)

git-svn-id: http://svn.digium.com/svn/asterisk/trunk@2239 f38db490-d61c-443f-a65b-d21fe96a405b

4 files changed, 979 insertions, 1 deletions

diff --git a/codecs/Makefile b/codecs/Makefile
index 742739af4..43c4e13db 100755
--- a/codecs/Makefile
+++ b/codecs/Makefile
@@ -34,7 +34,8 @@ LIBSPEEX+=-lspeex -lm
 LIBILBC=ilbc/libilbc.a
 
 CODECS+=$(MODG723) $(MODSPEEX) $(MODILBC) codec_gsm.so codec_lpc10.so  \
-        codec_adpcm.so codec_ulaw.so codec_alaw.so codec_a_mu.so
+        codec_adpcm.so codec_ulaw.so codec_alaw.so codec_a_mu.so \
+	codec_g726.so
 
 all: depend $(CODECS)
 
diff --git a/codecs/codec_g726.c b/codecs/codec_g726.c
new file mode 100755
index 000000000..ac32678ca
--- /dev/null
+++ b/codecs/codec_g726.c
@@ -0,0 +1,927 @@
+/* codec_g726.c - translate between signed linear and ITU G.726-32kbps
+ * 
+ * Asterisk -- A telephony toolkit for Linux.
+ *
+ * Based on frompcm.c and topcm.c from the Emiliano MIPL browser/
+ * interpreter.  See http://www.bsdtelephony.com.mx
+ *
+ * Copyright (c) 2004, Digium
+ *
+ * Mark Spencer <markster@digium.com>
+ *
+ * This program is free software, distributed under the terms of
+ * the GNU General Public License
+ */
+
+#include <asterisk/lock.h>
+#include <asterisk/logger.h>
+#include <asterisk/module.h>
+#include <asterisk/translate.h>
+#include <asterisk/channel.h>
+#include <fcntl.h>
+#include <netinet/in.h>
+#include <pthread.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+#define BUFFER_SIZE   8096	/* size for the translation buffers */
+#define BUF_SHIFT	5
+
+static ast_mutex_t localuser_lock = AST_MUTEX_INITIALIZER;
+static int localusecnt = 0;
+
+static char *tdesc = "ITU G.726-32kbps G726 Transcoder";
+
+/* Sample frame data */
+
+#include "slin_g726_ex.h"
+#include "g726_slin_ex.h"
+
+/*
+ * The following is the definition of the state structure
+ * used by the G.721/G.723 encoder and decoder to preserve their internal
+ * state between successive calls.  The meanings of the majority
+ * of the state structure fields are explained in detail in the
+ * CCITT Recommendation G.721.  The field names are essentially indentical
+ * to variable names in the bit level description of the coding algorithm
+ * included in this Recommendation.
+ */
+struct g726_state {
+	long yl;	/* Locked or steady state step size multiplier. */
+	short yu;	/* Unlocked or non-steady state step size multiplier. */
+	short dms;	/* Short term energy estimate. */
+	short dml;	/* Long term energy estimate. */
+	short ap;	/* Linear weighting coefficient of 'yl' and 'yu'. */
+
+	short a[2];	/* Coefficients of pole portion of prediction filter. */
+	short b[6];	/* Coefficients of zero portion of prediction filter. */
+	short pk[2];	/*
+			 * Signs of previous two samples of a partially
+			 * reconstructed signal.
+			 */
+	short dq[6];	/*
+			 * Previous 6 samples of the quantized difference
+			 * signal represented in an internal floating point
+			 * format.
+			 */
+	short sr[2];	/*
+			 * Previous 2 samples of the quantized difference
+			 * signal represented in an internal floating point
+			 * format.
+			 */
+	char td;	/* delayed tone detect, new in 1988 version */
+};
+
+
+
+static short qtab_721[7] = {-124, 80, 178, 246, 300, 349, 400};
+/*
+ * Maps G.721 code word to reconstructed scale factor normalized log
+ * magnitude values.
+ */
+static short	_dqlntab[16] = {-2048, 4, 135, 213, 273, 323, 373, 425,
+				425, 373, 323, 273, 213, 135, 4, -2048};
+
+/* Maps G.721 code word to log of scale factor multiplier. */
+static short	_witab[16] = {-12, 18, 41, 64, 112, 198, 355, 1122,
+				1122, 355, 198, 112, 64, 41, 18, -12};
+/*
+ * Maps G.721 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 short	_fitab[16] = {0, 0, 0, 0x200, 0x200, 0x200, 0x600, 0xE00,
+				0xE00, 0x600, 0x200, 0x200, 0x200, 0, 0, 0};
+
+static short power2[15] = {1, 2, 4, 8, 0x10, 0x20, 0x40, 0x80,
+			0x100, 0x200, 0x400, 0x800, 0x1000, 0x2000, 0x4000};
+
+/*
+ * quan()
+ *
+ * quantizes the input val against the table of size short integers.
+ * It returns i if table[i - 1] <= val < table[i].
+ *
+ * Using linear search for simple coding.
+ */
+static int quan(int val, short *table, int size)
+{
+	int		i;
+
+	for (i = 0; i < size; i++)
+		if (val < *table++)
+			break;
+	return (i);
+}
+
+/*
+ * fmult()
+ *
+ * returns the integer product of the 14-bit integer "an" and
+ * "floating point" representation (4-bit exponent, 6-bit mantessa) "srn".
+ */
+static int fmult(int an, int srn)
+{
+	short		anmag, anexp, anmant;
+	short		wanexp, wanmant;
+	short		retval;
+
+	anmag = (an > 0) ? an : ((-an) & 0x1FFF);
+	anexp = quan(anmag, power2, 15) - 6;
+	anmant = (anmag == 0) ? 32 :
+	    (anexp >= 0) ? anmag >> anexp : anmag << -anexp;
+	wanexp = anexp + ((srn >> 6) & 0xF) - 13;
+
+	wanmant = (anmant * (srn & 077) + 0x30) >> 4;
+	retval = (wanexp >= 0) ? ((wanmant << wanexp) & 0x7FFF) :
+	    (wanmant >> -wanexp);
+
+	return (((an ^ srn) < 0) ? -retval : retval);
+}
+
+/*
+ * g72x_init_state()
+ *
+ * This routine initializes and/or resets the g726_state structure
+ * pointed to by 'state_ptr'.
+ * All the initial state values are specified in the CCITT G.721 document.
+ */
+static void g726_init_state(struct g726_state *state_ptr)
+{
+	int		cnta;
+
+	state_ptr->yl = 34816;
+	state_ptr->yu = 544;
+	state_ptr->dms = 0;
+	state_ptr->dml = 0;
+	state_ptr->ap = 0;
+	for (cnta = 0; cnta < 2; cnta++) {
+		state_ptr->a[cnta] = 0;
+		state_ptr->pk[cnta] = 0;
+		state_ptr->sr[cnta] = 32;
+	}
+	for (cnta = 0; cnta < 6; cnta++) {
+		state_ptr->b[cnta] = 0;
+		state_ptr->dq[cnta] = 32;
+	}
+	state_ptr->td = 0;
+}
+
+/*
+ * predictor_zero()
+ *
+ * computes the estimated signal from 6-zero predictor.
+ *
+ */
+static int predictor_zero(struct g726_state *state_ptr)
+{
+	int		i;
+	int		sezi;
+
+	sezi = fmult(state_ptr->b[0] >> 2, state_ptr->dq[0]);
+	for (i = 1; i < 6; i++)			/* ACCUM */
+		sezi += fmult(state_ptr->b[i] >> 2, state_ptr->dq[i]);
+	return (sezi);
+}
+/*
+ * predictor_pole()
+ *
+ * computes the estimated signal from 2-pole predictor.
+ *
+ */
+static int predictor_pole(struct g726_state *state_ptr)
+{
+	return (fmult(state_ptr->a[1] >> 2, state_ptr->sr[1]) +
+	    fmult(state_ptr->a[0] >> 2, state_ptr->sr[0]));
+}
+
+/*
+ * step_size()
+ *
+ * computes the quantization step size of the adaptive quantizer.
+ *
+ */
+static int step_size(struct g726_state *state_ptr)
+{
+	int		y;
+	int		dif;
+	int		al;
+
+	if (state_ptr->ap >= 256)
+		return (state_ptr->yu);
+	else {
+		y = state_ptr->yl >> 6;
+		dif = state_ptr->yu - y;
+		al = state_ptr->ap >> 2;
+		if (dif > 0)
+			y += (dif * al) >> 6;
+		else if (dif < 0)
+			y += (dif * al + 0x3F) >> 6;
+		return (y);
+	}
+}
+
+/*
+ * quantize()
+ *
+ * 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 int quantize(
+	int		d,	/* Raw difference signal sample */
+	int		y,	/* Step size multiplier */
+	short		*table,	/* quantization table */
+	int		size)	/* table size of short integers */
+{
+	short		dqm;	/* Magnitude of 'd' */
+	short		exp;	/* Integer part of base 2 log of 'd' */
+	short		mant;	/* Fractional part of base 2 log */
+	short		dl;	/* Log of magnitude of 'd' */
+	short		dln;	/* Step size scale factor normalized log */
+	int		i;
+
+	/*
+	 * LOG
+	 *
+	 * Compute base 2 log of 'd', and store in 'dl'.
+	 */
+	dqm = abs(d);
+	exp = quan(dqm >> 1, power2, 15);
+	mant = ((dqm << 7) >> exp) & 0x7F;	/* Fractional portion. */
+	dl = (exp << 7) + mant;
+
+	/*
+	 * SUBTB
+	 *
+	 * "Divide" by step size multiplier.
+	 */
+	dln = dl - (y >> 2);
+
+	/*
+	 * QUAN
+	 *
+	 * Obtain codword i for 'd'.
+	 */
+	i = quan(dln, table, size);
+	if (d < 0)			/* take 1's complement of i */
+		return ((size << 1) + 1 - i);
+	else if (i == 0)		/* take 1's complement of 0 */
+		return ((size << 1) + 1); /* new in 1988 */
+	else
+		return (i);
+}
+
+/*
+ * reconstruct()
+ *
+ * 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 int reconstruct(
+	int		sign,	/* 0 for non-negative value */
+	int		dqln,	/* G.72x codeword */
+	int		y)	/* Step size multiplier */
+{
+	short		dql;	/* Log of 'dq' magnitude */
+	short		dex;	/* Integer part of log */
+	short		dqt;
+	short		dq;	/* Reconstructed difference signal sample */
+
+	dql = dqln + (y >> 2);	/* ADDA */
+
+	if (dql < 0) {
+		return ((sign) ? -0x8000 : 0);
+	} else {		/* ANTILOG */
+		dex = (dql >> 7) & 15;
+		dqt = 128 + (dql & 127);
+		dq = (dqt << 7) >> (14 - dex);
+		return ((sign) ? (dq - 0x8000) : dq);
+	}
+}
+
+/*
+ * update()
+ *
+ * updates the state variables for each output code
+ */
+static void update(
+	int		code_size,	/* distinguish 723_40 with others */
+	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 */
+	struct g726_state *state_ptr)	/* coder state pointer */
+{
+	int		cnt;
+	short		mag, exp;	/* Adaptive predictor, FLOAT A */
+	short		a2p=0;		/* LIMC */
+	short		a1ul;		/* UPA1 */
+	short		pks1;	/* UPA2 */
+	short		fa1;
+	char		tr;		/* tone/transition detector */
+	short		ylint, thr2, dqthr;
+	short  		ylfrac, thr1;
+	short		pk0;
+
+	pk0 = (dqsez < 0) ? 1 : 0;	/* needed in updating predictor poles */
+
+	mag = dq & 0x7FFF;		/* prediction difference magnitude */
+	/* TRANS */
+	ylint = state_ptr->yl >> 15;	/* exponent part of yl */
+	ylfrac = (state_ptr->yl >> 10) & 0x1F;	/* fractional part of yl */
+	thr1 = (32 + ylfrac) << ylint;		/* threshold */
+	thr2 = (ylint > 9) ? 31 << 10 : thr1;	/* limit thr2 to 31 << 10 */
+	dqthr = (thr2 + (thr2 >> 1)) >> 1;	/* dqthr = 0.75 * thr2 */
+	if (state_ptr->td == 0)		/* signal supposed voice */
+		tr = 0;
+	else if (mag <= dqthr)		/* supposed data, but small mag */
+		tr = 0;			/* treated as voice */
+	else				/* signal is data (modem) */
+		tr = 1;
+
+	/*
+	 * Quantizer scale factor adaptation.
+	 */
+
+	/* FUNCTW & FILTD & DELAY */
+	/* update non-steady state step size multiplier */
+	state_ptr->yu = y + ((wi - y) >> 5);
+
+	/* LIMB */
+	if (state_ptr->yu < 544)	/* 544 <= yu <= 5120 */
+		state_ptr->yu = 544;
+	else if (state_ptr->yu > 5120)
+		state_ptr->yu = 5120;
+
+	/* FILTE & DELAY */
+	/* update steady state step size multiplier */
+	state_ptr->yl += state_ptr->yu + ((-state_ptr->yl) >> 6);
+
+	/*
+	 * Adaptive predictor coefficients.
+	 */
+	if (tr == 1) {			/* reset a's and b's for modem signal */
+		state_ptr->a[0] = 0;
+		state_ptr->a[1] = 0;
+		state_ptr->b[0] = 0;
+		state_ptr->b[1] = 0;
+		state_ptr->b[2] = 0;
+		state_ptr->b[3] = 0;
+		state_ptr->b[4] = 0;
+		state_ptr->b[5] = 0;
+	} else {			/* update a's and b's */
+		pks1 = pk0 ^ state_ptr->pk[0];		/* UPA2 */
+
+		/* update predictor pole a[1] */
+		a2p = state_ptr->a[1] - (state_ptr->a[1] >> 7);
+		if (dqsez != 0) {
+			fa1 = (pks1) ? state_ptr->a[0] : -state_ptr->a[0];
+			if (fa1 < -8191)	/* a2p = function of fa1 */
+				a2p -= 0x100;
+			else if (fa1 > 8191)
+				a2p += 0xFF;
+			else
+				a2p += fa1 >> 5;
+
+			if (pk0 ^ state_ptr->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 */
+		state_ptr->a[1] = a2p;
+
+		/* UPA1 */
+		/* update predictor pole a[0] */
+		state_ptr->a[0] -= state_ptr->a[0] >> 8;
+		if (dqsez != 0) {
+			if (pks1 == 0)
+				state_ptr->a[0] += 192;
+			else
+				state_ptr->a[0] -= 192;
+		}
+		/* LIMD */
+		a1ul = 15360 - a2p;
+		if (state_ptr->a[0] < -a1ul)
+			state_ptr->a[0] = -a1ul;
+		else if (state_ptr->a[0] > a1ul)
+			state_ptr->a[0] = a1ul;
+
+		/* UPB : update predictor zeros b[6] */
+		for (cnt = 0; cnt < 6; cnt++) {
+			if (code_size == 5)		/* for 40Kbps G.723 */
+				state_ptr->b[cnt] -= state_ptr->b[cnt] >> 9;
+			else			/* for G.721 and 24Kbps G.723 */
+				state_ptr->b[cnt] -= state_ptr->b[cnt] >> 8;
+			if (dq & 0x7FFF) {			/* XOR */
+				if ((dq ^ state_ptr->dq[cnt]) >= 0)
+					state_ptr->b[cnt] += 128;
+				else
+					state_ptr->b[cnt] -= 128;
+			}
+		}
+	}
+
+	for (cnt = 5; cnt > 0; cnt--)
+		state_ptr->dq[cnt] = state_ptr->dq[cnt-1];
+	/* FLOAT A : convert dq[0] to 4-bit exp, 6-bit mantissa f.p. */
+	if (mag == 0) {
+		state_ptr->dq[0] = (dq >= 0) ? 0x20 : 0xFC20;
+	} else {
+		exp = quan(mag, power2, 15);
+		state_ptr->dq[0] = (dq >= 0) ?
+		    (exp << 6) + ((mag << 6) >> exp) :
+		    (exp << 6) + ((mag << 6) >> exp) - 0x400;
+	}
+
+	state_ptr->sr[1] = state_ptr->sr[0];
+	/* FLOAT B : convert sr to 4-bit exp., 6-bit mantissa f.p. */
+	if (sr == 0) {
+		state_ptr->sr[0] = 0x20;
+	} else if (sr > 0) {
+		exp = quan(sr, power2, 15);
+		state_ptr->sr[0] = (exp << 6) + ((sr << 6) >> exp);
+	} else if (sr > -32768) {
+		mag = -sr;
+		exp = quan(mag, power2, 15);
+		state_ptr->sr[0] =  (exp << 6) + ((mag << 6) >> exp) - 0x400;
+	} else
+		state_ptr->sr[0] = 0xFC20;
+
+	/* DELAY A */
+	state_ptr->pk[1] = state_ptr->pk[0];
+	state_ptr->pk[0] = pk0;
+
+	/* TONE */
+	if (tr == 1)		/* this sample has been treated as data */
+		state_ptr->td = 0;	/* next one will be treated as voice */
+	else if (a2p < -11776)	/* small sample-to-sample correlation */
+		state_ptr->td = 1;	/* signal may be data */
+	else				/* signal is voice */
+		state_ptr->td = 0;
+
+	/*
+	 * Adaptation speed control.
+	 */
+	state_ptr->dms += (fi - state_ptr->dms) >> 5;		/* FILTA */
+	state_ptr->dml += (((fi << 2) - state_ptr->dml) >> 7);	/* FILTB */
+
+	if (tr == 1)
+		state_ptr->ap = 256;
+	else if (y < 1536)					/* SUBTC */
+		state_ptr->ap += (0x200 - state_ptr->ap) >> 4;
+	else if (state_ptr->td == 1)
+		state_ptr->ap += (0x200 - state_ptr->ap) >> 4;
+	else if (abs((state_ptr->dms << 2) - state_ptr->dml) >=
+	    (state_ptr->dml >> 3))
+		state_ptr->ap += (0x200 - state_ptr->ap) >> 4;
+	else
+		state_ptr->ap += (-state_ptr->ap) >> 4;
+}
+
+/*
+ * g726_decode()
+ *
+ * Description:
+ *
+ * Decodes a 4-bit code of G.726-32 encoded data of i and
+ * returns the resulting linear PCM, A-law or u-law value.
+ * return -1 for unknown out_coding value.
+ */
+static int g726_decode(int	i, struct g726_state *state_ptr)
+{
+	short		sezi, sei, sez, se;	/* ACCUM */
+	short		y;			/* MIX */
+	short		sr;			/* ADDB */
+	short		dq;
+	short		dqsez;
+
+	i &= 0x0f;			/* mask to get proper bits */
+	sezi = predictor_zero(state_ptr);
+	sez = sezi >> 1;
+	sei = sezi + predictor_pole(state_ptr);
+	se = sei >> 1;			/* se = estimated signal */
+
+	y = step_size(state_ptr);	/* dynamic quantizer step size */
+
+	dq = reconstruct(i & 0x08, _dqlntab[i], y); /* quantized diff. */
+
+	sr = (dq < 0) ? (se - (dq & 0x3FFF)) : se + dq;	/* reconst. signal */
+
+	dqsez = sr - se + sez;			/* pole prediction diff. */
+
+	update(4, y, _witab[i] << 5, _fitab[i], dq, sr, dqsez, state_ptr);
+
+	return (sr << 2);	/* sr was 14-bit dynamic range */
+}
+/*
+ * g726_encode()
+ *
+ * Encodes the input vale of linear PCM, A-law or u-law data sl and returns
+ * the resulting code. -1 is returned for unknown input coding value.
+ */
+static int g726_encode(int sl, struct g726_state *state_ptr)
+{
+	short		sezi, se, sez;		/* ACCUM */
+	short		d;			/* SUBTA */
+	short		sr;			/* ADDB */
+	short		y;			/* MIX */
+	short		dqsez;			/* ADDC */
+	short		dq, i;
+
+	sl >>= 2;			/* 14-bit dynamic range */
+
+	sezi = predictor_zero(state_ptr);
+	sez = sezi >> 1;
+	se = (sezi + predictor_pole(state_ptr)) >> 1;	/* estimated signal */
+
+	d = sl - se;				/* estimation difference */
+
+	/* quantize the prediction difference */
+	y = step_size(state_ptr);		/* quantizer step size */
+	i = quantize(d, y, qtab_721, 7);	/* i = G726 code */
+
+	dq = reconstruct(i & 8, _dqlntab[i], y);	/* quantized est diff */
+
+	sr = (dq < 0) ? se - (dq & 0x3FFF) : se + dq;	/* reconst. signal */
+
+	dqsez = sr + sez - se;			/* pole prediction diff. */
+
+	update(4, y, _witab[i] << 5, _fitab[i], dq, sr, dqsez, state_ptr);
+
+	return (i);
+}
+
+/*
+ * Private workspace for translating signed linear signals to G726.
+ */
+
+struct g726_encoder_pvt
+{
+  struct ast_frame f;
+  char offset[AST_FRIENDLY_OFFSET];   /* Space to build offset */
+  unsigned char outbuf[BUFFER_SIZE];  /* Encoded G726, two nibbles to a word */
+  unsigned char next_flag;
+  struct g726_state g726;
+  int tail;
+};
+
+/*
+ * Private workspace for translating G726 signals to signed linear.
+ */
+
+struct g726_decoder_pvt
+{
+  struct ast_frame f;
+  char offset[AST_FRIENDLY_OFFSET];	/* Space to build offset */
+  short outbuf[BUFFER_SIZE];	/* Decoded signed linear values */
+  struct g726_state g726;
+  int tail;
+};
+
+/*
+ * G726ToLin_New
+ *  Create a new instance of g726_decoder_pvt.
+ *
+ * Results:
+ *  Returns a pointer to the new instance.
+ *
+ * Side effects:
+ *  None.
+ */
+
+static struct ast_translator_pvt *
+g726tolin_new (void)
+{
+  struct g726_decoder_pvt *tmp;
+  tmp = malloc (sizeof (struct g726_decoder_pvt));
+  if (tmp)
+    {
+	  memset(tmp, 0, sizeof(*tmp));
+      tmp->tail = 0;
+      localusecnt++;
+	  g726_init_state(&tmp->g726);
+      ast_update_use_count ();
+    }
+  return (struct ast_translator_pvt *) tmp;
+}
+
+/*
+ * LinToG726_New
+ *  Create a new instance of g726_encoder_pvt.
+ *
+ * Results:
+ *  Returns a pointer to the new instance.
+ *
+ * Side effects:
+ *  None.
+ */
+
+static struct ast_translator_pvt *
+lintog726_new (void)
+{
+  struct g726_encoder_pvt *tmp;
+  tmp = malloc (sizeof (struct g726_encoder_pvt));
+  if (tmp)
+    {
+	  memset(tmp, 0, sizeof(*tmp));
+      localusecnt++;
+      tmp->tail = 0;
+	  g726_init_state(&tmp->g726);
+      ast_update_use_count ();
+    }
+  return (struct ast_translator_pvt *) tmp;
+}
+
+/*
+ * G726ToLin_FrameIn
+ *  Fill an input buffer with packed 4-bit G726 values if there is room
+ *  left.
+ *
+ * Results:
+ *  Foo
+ *
+ * Side effects:
+ *  tmp->tail is the number of packed values in the buffer.
+ */
+
+static int
+g726tolin_framein (struct ast_translator_pvt *pvt, struct ast_frame *f)
+{
+  struct g726_decoder_pvt *tmp = (struct g726_decoder_pvt *) pvt;
+  unsigned char *b;
+  int x;
+
+  b = f->data;
+  for (x=0;x<f->datalen;x++) {
+  	if (tmp->tail >= BUFFER_SIZE) {
+		ast_log(LOG_WARNING, "Out of buffer space!\n");
+		return -1;
+	}
+	tmp->outbuf[tmp->tail++] = g726_decode((b[x] >> 4) & 0xf, &tmp->g726);
+  	if (tmp->tail >= BUFFER_SIZE) {
+		ast_log(LOG_WARNING, "Out of buffer space!\n");
+		return -1;
+	}
+	tmp->outbuf[tmp->tail++] = g726_decode(b[x] & 0x0f, &tmp->g726);
+  }
+
+  return 0;
+}
+
+/*
+ * G726ToLin_FrameOut
+ *  Convert 4-bit G726 encoded signals to 16-bit signed linear.
+ *
+ * Results:
+ *  Converted signals are placed in tmp->f.data, tmp->f.datalen
+ *  and tmp->f.samples are calculated.
+ *
+ * Side effects:
+ *  None.
+ */
+
+static struct ast_frame *
+g726tolin_frameout (struct ast_translator_pvt *pvt)
+{
+  struct g726_decoder_pvt *tmp = (struct g726_decoder_pvt *) pvt;
+
+  if (!tmp->tail)
+    return NULL;
+
+  tmp->f.frametype = AST_FRAME_VOICE;
+  tmp->f.subclass = AST_FORMAT_SLINEAR;
+  tmp->f.datalen = tmp->tail * 2;
+  tmp->f.samples = tmp->tail;
+  tmp->f.mallocd = 0;
+  tmp->f.offset = AST_FRIENDLY_OFFSET;
+  tmp->f.src = __PRETTY_FUNCTION__;
+  tmp->f.data = tmp->outbuf;
+  tmp->tail = 0;
+  return &tmp->f;
+}
+
+/*
+ * LinToG726_FrameIn
+ *  Fill an input buffer with 16-bit signed linear PCM values.
+ *
+ * Results:
+ *  None.
+ *
+ * Side effects:
+ *  tmp->tail is number of signal values in the input buffer.
+ */
+
+static int
+lintog726_framein (struct ast_translator_pvt *pvt, struct ast_frame *f)
+{
+  struct g726_encoder_pvt *tmp = (struct g726_encoder_pvt *) pvt;
+  short *s = f->data;
+  int samples = f->datalen / 2;
+  int x;
+  for (x=0;x<samples;x++) {
+  	if (tmp->next_flag & 0x80) {
+		if (tmp->tail >= BUFFER_SIZE) {
+			ast_log(LOG_WARNING, "Out of buffer space\n");
+			return -1;
+		}
+		tmp->outbuf[tmp->tail++] = ((tmp->next_flag & 0xf)<< 4) | g726_encode(s[x], &tmp->g726);
+		tmp->next_flag = 0;
+	} else {
+		tmp->next_flag = 0x80 | g726_encode(s[x], &tmp->g726);
+	}
+  }
+  return 0;
+}
+
+/*
+ * LinToG726_FrameOut
+ *  Convert a buffer of raw 16-bit signed linear PCM to a buffer
+ *  of 4-bit G726 packed two to a byte (Big Endian).
+ *
+ * Results:
+ *  Foo
+ *
+ * Side effects:
+ *  Leftover inbuf data gets packed, tail gets updated.
+ */
+
+static struct ast_frame *
+lintog726_frameout (struct ast_translator_pvt *pvt)
+{
+  struct g726_encoder_pvt *tmp = (struct g726_encoder_pvt *) pvt;
+  
+  if (!tmp->tail)
+  	return NULL;
+  tmp->f.frametype = AST_FRAME_VOICE;
+  tmp->f.subclass = AST_FORMAT_G726;
+  tmp->f.samples = tmp->tail * 2;
+  tmp->f.mallocd = 0;
+  tmp->f.offset = AST_FRIENDLY_OFFSET;
+  tmp->f.src = __PRETTY_FUNCTION__;
+  tmp->f.data = tmp->outbuf;
+  tmp->f.datalen = tmp->tail;
+
+  tmp->tail = 0;
+  return &tmp->f;
+}
+
+
+/*
+ * G726ToLin_Sample
+ */
+
+static struct ast_frame *
+g726tolin_sample (void)
+{
+  static struct ast_frame f;
+  f.frametype = AST_FRAME_VOICE;
+  f.subclass = AST_FORMAT_G726;
+  f.datalen = sizeof (g726_slin_ex);
+  f.samples = sizeof(g726_slin_ex) * 2;
+  f.mallocd = 0;
+  f.offset = 0;
+  f.src = __PRETTY_FUNCTION__;
+  f.data = g726_slin_ex;
+  return &f;
+}
+
+/*
+ * LinToG726_Sample
+ */
+
+static struct ast_frame *
+lintog726_sample (void)
+{
+  static struct ast_frame f;
+  f.frametype = AST_FRAME_VOICE;
+  f.subclass = AST_FORMAT_SLINEAR;
+  f.datalen = sizeof (slin_g726_ex);
+  /* Assume 8000 Hz */
+  f.samples = sizeof (slin_g726_ex) / 2;
+  f.mallocd = 0;
+  f.offset = 0;
+  f.src = __PRETTY_FUNCTION__;
+  f.data = slin_g726_ex;
+  return &f;
+}
+
+/*
+ * G726_Destroy
+ *  Destroys a private workspace.
+ *
+ * Results:
+ *  It's gone!
+ *
+ * Side effects:
+ *  None.
+ */
+
+static void
+g726_destroy (struct ast_translator_pvt *pvt)
+{
+  free (pvt);
+  localusecnt--;
+  ast_update_use_count ();
+}
+
+/*
+ * The complete translator for G726ToLin.
+ */
+
+static struct ast_translator g726tolin = {
+  "g726tolin",
+  AST_FORMAT_G726,
+  AST_FORMAT_SLINEAR,
+  g726tolin_new,
+  g726tolin_framein,
+  g726tolin_frameout,
+  g726_destroy,
+  /* NULL */
+  g726tolin_sample
+};
+
+/*
+ * The complete translator for LinToG726.
+ */
+
+static struct ast_translator lintog726 = {
+  "lintog726",
+  AST_FORMAT_SLINEAR,
+  AST_FORMAT_G726,
+  lintog726_new,
+  lintog726_framein,
+  lintog726_frameout,
+  g726_destroy,
+  /* NULL */
+  lintog726_sample
+};
+
+int
+unload_module (void)
+{
+  int res;
+  ast_mutex_lock (&localuser_lock);
+  res = ast_unregister_translator (&lintog726);
+  if (!res)
+    res = ast_unregister_translator (&g726tolin);
+  if (localusecnt)
+    res = -1;
+  ast_mutex_unlock (&localuser_lock);
+  return res;
+}
+
+int
+load_module (void)
+{
+  int res;
+  res = ast_register_translator (&g726tolin);
+  if (!res)
+    res = ast_register_translator (&lintog726);
+  else
+    ast_unregister_translator (&g726tolin);
+  return res;
+}
+
+/*
+ * Return a description of this module.
+ */
+
+char *
+description (void)
+{
+  return tdesc;
+}
+
+int
+usecount (void)
+{
+  int res;
+  STANDARD_USECOUNT (res);
+  return res;
+}
+
+char *
+key ()
+{
+  return ASTERISK_GPL_KEY;
+}
diff --git a/codecs/g726_slin_ex.h b/codecs/g726_slin_ex.h
new file mode 100755
index 000000000..b5bff5f80
--- /dev/null
+++ b/codecs/g726_slin_ex.h
@@ -0,0 +1,25 @@
+/*
+ * adpcm_slin_ex.h --
+ *
+ *	4-bit G.726 data, 20 milliseconds worth at 8 kHz.
+ *
+ * Source: g726.example
+ *
+ * Copyright (C) 2001, Linux Support Services, Inc.
+ *
+ * Distributed under the terms of the GNU General Public License
+ *
+ */
+
+static unsigned char g726_slin_ex[] = {
+	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
+};
diff --git a/codecs/slin_g726_ex.h b/codecs/slin_g726_ex.h
new file mode 100755
index 000000000..efd516a11
--- /dev/null
+++ b/codecs/slin_g726_ex.h
@@ -0,0 +1,25 @@
+/*
+ * slin_adpcm_ex.h --
+ *
+ *	Signed 16-bit audio data, 10 milliseconds worth at 8 kHz.
+ *
+ * Source: g726.example
+ *
+ * Copyright (C) 2001, Linux Support Services, Inc.
+ *
+ * Distributed under the terms of the GNU General Public License
+ *
+ */
+
+static signed short slin_g726_ex[] = {
+	0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
+	0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
+	0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
+	0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
+	0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
+	0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
+	0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
+	0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
+	0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000,
+	0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000
+};