mirror of https://github.com/sipwise/sems.git
git-svn-id: http://svn.berlios.de/svnroot/repos/sems/trunk@596 8eb893ce-cfd4-0310-b710-fb5ebe64c474sayer/1.4-spce2.6
Stefan Sayer
19 years ago
parent
1e2c65a43d
commit
8922409cf1
@ -0,0 +1,32 @@
|
||||
plug_in_name = speex
|
||||
|
||||
SPEEX_INC?=/usr/include/
|
||||
SPEEX_LIB?=-lspeex
|
||||
|
||||
# or:
|
||||
# SPEEXDIR?=speex_src
|
||||
# SPEEX_INC?=$(SPEEXDIR)
|
||||
# SPEEX_LIB?=$(SPEEXDIR)/libspeex/.libs/libspeex.a
|
||||
#
|
||||
# CONFIGURE_FLAG=--with-pic
|
||||
|
||||
|
||||
module_ldflags = $(SPEEX_LIB)
|
||||
module_cflags = -I $(SPEEX_INC) -ansi # -DNOFPU
|
||||
|
||||
# or
|
||||
# module_extra_objs = $(SPEEX_LIB)
|
||||
#extra_clean = clean_speex
|
||||
|
||||
include ../Makefile.audio_module
|
||||
|
||||
|
||||
#$(SPEEXDIR)/Makefile:
|
||||
# cd $(SPEEXDIR); ./configure $(CONFIGURE_FLAG)
|
||||
#
|
||||
#$(SPEEXDIR)/libspeex/.libs/libspeex.a:: $(SPEEXDIR)/Makefile
|
||||
# $(MAKE) -C $(SPEEXDIR)/libspeex
|
||||
#
|
||||
#.PHONY: clean_speex
|
||||
#clean_speex:
|
||||
# $(MAKE) -C $(SPEEXDIR) distclean
|
||||
@ -0,0 +1,381 @@
|
||||
/*
|
||||
This is a simple interface to the Speex library for Sems.
|
||||
Copyright (C) 2005 Ettore Benedetti e.benedetti@elitel.biz
|
||||
updated 2007 Stefan Sayer sayer@iptel.org
|
||||
|
||||
This code is free software; you can redistribute it and/or
|
||||
modify it under the terms of the GNU Lesser General Public
|
||||
License as published by the Free Software Foundation.
|
||||
|
||||
This code 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 library; if not, write to the Free Software
|
||||
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
||||
*/
|
||||
|
||||
#include <string.h>
|
||||
#include <stdlib.h>
|
||||
|
||||
#include "amci.h"
|
||||
#include "codecs.h"
|
||||
#include "speex/speex.h"
|
||||
#include "../../log.h"
|
||||
|
||||
/* Speex constants */
|
||||
#define SPEEX_FRAME_MS 20
|
||||
#define SPEEX_NB_SAMPLES_PER_FRAME 160
|
||||
#define SPEEX_WB_SAMPLES_PER_FRAME 320
|
||||
|
||||
/* Default encoder settings */
|
||||
#define MODE 5
|
||||
#define FRAMES_PER_PACKET 1 /* or other implementations choke */
|
||||
#define BYTES_PER_FRAME (160 / 8) /* depends on mode */
|
||||
|
||||
/* #ifdef NOFPU */
|
||||
/* #warning Using integer encoding/decoding algorithms */
|
||||
/* #else */
|
||||
/* #warning Using floating point encoding/decoding algorithms */
|
||||
/* #endif */
|
||||
|
||||
/*
|
||||
Important: Frame length is always 20ms!
|
||||
|
||||
Narrow-band mode (8khz)
|
||||
|
||||
Mode Bit-rate(bps) Frame size(bits) Mips Notes
|
||||
0 250 5 N/A No transmission (DTX)
|
||||
1 2150 43 6 Vocoder (mostly for comfort noise)
|
||||
2 5950 119 9 Very noticeable artifacts/noise, good intelligibility
|
||||
3 8000 160 10 Artifacts/noise sometimes noticeable
|
||||
4 11000 220 14 Artifacts usually noticeable only with headphones
|
||||
5 15000 300 11 Need good headphones to tell the difference
|
||||
6 18200 364 17.5 Hard to tell the difference even with good headphones
|
||||
7 24600 492 14.5 Completely transparent for voice, good quality music
|
||||
8 3950 79 10.5 Very noticeable artifacts/noise, good intelligibility
|
||||
|
||||
Wide-band mode (16khz). A wide-band packet is composed
|
||||
by a narrow-band packet plus an extension, described here.
|
||||
The two band components are created by means of a QMF.
|
||||
|
||||
Mode Additional bit-rate(bps) Additional frame size(bits)
|
||||
0 +200 +4
|
||||
1 +1800 +36
|
||||
2 +5600 +112
|
||||
3 +9600 +192
|
||||
4 +17600 +352
|
||||
*/
|
||||
|
||||
int Pcm16_2_SpeexNB( unsigned char* out_buf, unsigned char* in_buf, unsigned int size,
|
||||
unsigned int channels, unsigned int rate, long h_codec );
|
||||
int SpeexNB_2_Pcm16( unsigned char* out_buf, unsigned char* in_buf, unsigned int size,
|
||||
unsigned int channels, unsigned int rate, long h_codec );
|
||||
|
||||
long speexNB_create(const char* format_parameters, amci_codec_fmt_info_t* format_description);
|
||||
void speexNB_destroy(long handle);
|
||||
|
||||
static unsigned int speexNB_bytes2samples(long, unsigned int);
|
||||
static unsigned int speexNB_samples2bytes(long, unsigned int);
|
||||
|
||||
BEGIN_EXPORTS("speex")
|
||||
|
||||
BEGIN_CODECS
|
||||
CODEC(CODEC_SPEEX_NB, Pcm16_2_SpeexNB, SpeexNB_2_Pcm16, NULL, speexNB_create, speexNB_destroy,
|
||||
speexNB_bytes2samples, speexNB_samples2bytes)
|
||||
END_CODECS
|
||||
|
||||
BEGIN_PAYLOADS
|
||||
PAYLOAD(-1, "speex", 8000, 1, CODEC_SPEEX_NB, AMCI_PT_AUDIO_FRAME)
|
||||
END_PAYLOADS
|
||||
|
||||
BEGIN_FILE_FORMATS
|
||||
END_FILE_FORMATS
|
||||
|
||||
END_EXPORTS
|
||||
|
||||
typedef struct {
|
||||
void *state;
|
||||
SpeexBits bits;
|
||||
#ifndef NOFPU
|
||||
float pool[AUDIO_BUFFER_SIZE];
|
||||
#endif
|
||||
} OneWay;
|
||||
|
||||
typedef struct {
|
||||
OneWay *encoder;
|
||||
OneWay *decoder;
|
||||
|
||||
/* Encoder settings */
|
||||
int frames_per_packet;
|
||||
int mode; /* 0..15 */
|
||||
|
||||
/* Decoder settings */
|
||||
int perceptual; /* 0,1 */
|
||||
|
||||
} SpeexState;
|
||||
|
||||
/* See table, above */
|
||||
static const int nb_encoded_frame_bits[] = { 5, 43, 119, 160, 220, 300, 364, 492, 79 };
|
||||
|
||||
/*
|
||||
Search for a parameter assignement in input string.
|
||||
If it's not found *param_value is null, otherwise *param_value points to the right hand term.
|
||||
In both cases a pointer suitable for a new search is returned
|
||||
*/
|
||||
static char* read_param(char* input, const char *param, char** param_value)
|
||||
{
|
||||
int param_size;
|
||||
|
||||
/* Eat spaces and semi-colons */
|
||||
while (*input==' ' && *input==';')
|
||||
input++;
|
||||
|
||||
*param_value = NULL;
|
||||
param_size = strlen(param);
|
||||
if (strncmp(input, param, param_size))
|
||||
return input;
|
||||
if (*(input+param_size) != '=')
|
||||
return input;
|
||||
input+=param_size+1;
|
||||
|
||||
/* Found and discarded a matching parameter */
|
||||
*param_value = input;
|
||||
while (*input && *input!=' ' && *input!=';')
|
||||
input++;
|
||||
|
||||
if (*input)
|
||||
*input++ = 0;
|
||||
|
||||
return input;
|
||||
}
|
||||
|
||||
long speexNB_create(const char* format_parameters, amci_codec_fmt_info_t* format_description)
|
||||
{
|
||||
SpeexState* ss;
|
||||
const int BLEN = 63;
|
||||
int bits;
|
||||
|
||||
ss = (SpeexState*) calloc(1, sizeof(SpeexState));
|
||||
|
||||
if (!ss)
|
||||
return -1;
|
||||
|
||||
/* Note that
|
||||
1) SEMS ignore a=ptime: SDP parameter so we can choose the one we like
|
||||
2) Multiple frames in a packet don't need to be byte-aligned
|
||||
*/
|
||||
ss->frames_per_packet = FRAMES_PER_PACKET;
|
||||
|
||||
/* Compression mode (see table) */
|
||||
ss->mode = MODE;
|
||||
|
||||
/* Perceptual enhancement. Turned on by default */
|
||||
ss->perceptual = 1;
|
||||
|
||||
/* See draft-herlein-avt-rtp-speex-00.txt */
|
||||
if (format_parameters && strlen(format_parameters)<=BLEN){
|
||||
|
||||
char buffer2[BLEN+1];
|
||||
char *buffer = buffer2;
|
||||
|
||||
strcpy(buffer, format_parameters);
|
||||
|
||||
while (*buffer) {
|
||||
char* error;
|
||||
char *param_value;
|
||||
|
||||
/* Speex encoding mode (assume NB) */
|
||||
buffer=read_param(buffer, "mode", ¶m_value);
|
||||
if (param_value) {
|
||||
int mode;
|
||||
if (strcmp(param_value, "any")) {
|
||||
mode = strtol(param_value, &error, 10);
|
||||
if (!*error && mode>=0 && mode<=8)
|
||||
ss->mode = mode;
|
||||
}
|
||||
continue;
|
||||
}
|
||||
|
||||
/* Perceptual enhancement */
|
||||
buffer=read_param(buffer, "penh", ¶m_value);
|
||||
if (param_value) {
|
||||
if (!strcmp(param_value,"no"))
|
||||
ss->perceptual=0;
|
||||
continue;
|
||||
}
|
||||
|
||||
/* Unknown parameter */
|
||||
if (*buffer) {
|
||||
param_value = buffer;
|
||||
while (*buffer && *buffer!=';')
|
||||
buffer++;
|
||||
|
||||
if (*buffer)
|
||||
*buffer++ = 0;
|
||||
|
||||
WARN("SDP parameter fmtp: %s not set in speex.\n", param_value);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/* Meaningful bits in one packet */
|
||||
bits = nb_encoded_frame_bits[ss->mode] * ss->frames_per_packet;
|
||||
|
||||
format_description[0].id = AMCI_FMT_FRAME_LENGTH;
|
||||
format_description[0].value = SPEEX_FRAME_MS * ss->frames_per_packet;
|
||||
|
||||
format_description[1].id = AMCI_FMT_FRAME_SIZE;
|
||||
format_description[1].value = SPEEX_NB_SAMPLES_PER_FRAME * ss->frames_per_packet;
|
||||
|
||||
format_description[2].id = AMCI_FMT_ENCODED_FRAME_SIZE;
|
||||
format_description[2].value = bits/8 + (bits % 8 ? 1 : 0) + 1;
|
||||
|
||||
DBG("set AMCI_FMT_FRAME_LENGTH to %d\n", format_description[0].value);
|
||||
DBG("set AMCI_FMT_FRAME_SIZE to %d\n", format_description[1].value);
|
||||
DBG("set AMCI_FMT_ENCODED_FRAME_SIZE to %d\n", format_description[2].value);
|
||||
|
||||
format_description[3].id = 0;
|
||||
|
||||
DBG("SpeexState %p inserted with mode %d and %d frames per packet,\n", ss, ss->mode, ss->frames_per_packet);
|
||||
|
||||
return (long)ss;
|
||||
}
|
||||
|
||||
void speexNB_destroy(long handle)
|
||||
{
|
||||
SpeexState* ss = (SpeexState*) handle;
|
||||
|
||||
DBG("SpeexDestroy for handle %ld\n", handle);
|
||||
|
||||
if (!ss)
|
||||
return;
|
||||
|
||||
if (ss->encoder) {
|
||||
speex_encoder_destroy(ss->encoder->state);
|
||||
speex_bits_destroy(&ss->encoder->bits);
|
||||
free(ss->encoder);
|
||||
}
|
||||
|
||||
if (ss->decoder) {
|
||||
speex_decoder_destroy(ss->decoder->state);
|
||||
speex_bits_destroy(&ss->decoder->bits);
|
||||
free(ss->decoder);
|
||||
}
|
||||
|
||||
free(ss);
|
||||
}
|
||||
|
||||
static unsigned int speexNB_bytes2samples(long h_codec, unsigned int num_bytes) {
|
||||
return (SPEEX_NB_SAMPLES_PER_FRAME * num_bytes) / BYTES_PER_FRAME;
|
||||
}
|
||||
|
||||
static unsigned int speexNB_samples2bytes(long h_codec, unsigned int num_samples) {
|
||||
return BYTES_PER_FRAME * num_samples / SPEEX_NB_SAMPLES_PER_FRAME;
|
||||
}
|
||||
|
||||
|
||||
int Pcm16_2_SpeexNB( unsigned char* out_buf, unsigned char* in_buf, unsigned int size,
|
||||
unsigned int channels, unsigned int rate, long h_codec )
|
||||
{
|
||||
SpeexState* ss;
|
||||
OneWay *encoder;
|
||||
const short* pcm = (const short*) in_buf;
|
||||
char* buffer = (char*)out_buf;
|
||||
div_t blocks;
|
||||
|
||||
ss = (SpeexState*) h_codec;
|
||||
|
||||
if (!ss || channels!=1 || rate!=8000)
|
||||
return -1;
|
||||
|
||||
/* encoder init */
|
||||
if (!(encoder=ss->encoder)) {
|
||||
ss->encoder = encoder = (OneWay*) calloc(1, sizeof(OneWay));
|
||||
if (!encoder)
|
||||
return -1;
|
||||
encoder->state = speex_encoder_init(&speex_nb_mode);
|
||||
speex_bits_init(&encoder->bits);
|
||||
speex_encoder_ctl(encoder->state, SPEEX_SET_MODE, &ss->mode);
|
||||
}
|
||||
|
||||
blocks = div(size, sizeof(short)*SPEEX_NB_SAMPLES_PER_FRAME);
|
||||
if (blocks.rem) {
|
||||
ERROR("Pcm16_2_Speex: not integral number of blocks %d.%d\n", blocks.quot, blocks.rem);
|
||||
return -1;
|
||||
}
|
||||
|
||||
/* For each chunk of ss->frame_size bytes, encode a single frame */
|
||||
speex_bits_reset(&encoder->bits);
|
||||
while (blocks.quot--) {
|
||||
|
||||
#ifdef NOFPU
|
||||
speex_encode_int(encoder->state, pcm, &encoder->bits);
|
||||
pcm+=SPEEX_NB_SAMPLES_PER_FRAME;
|
||||
#else
|
||||
int i;
|
||||
for (i=0; i<SPEEX_NB_SAMPLES_PER_FRAME; i++)
|
||||
encoder->pool[i] = (float)*pcm++;
|
||||
speex_encode(encoder->state, encoder->pool, &encoder->bits);
|
||||
#endif
|
||||
}
|
||||
|
||||
buffer += speex_bits_write(&encoder->bits, buffer, AUDIO_BUFFER_SIZE);
|
||||
|
||||
return buffer - (char*)out_buf;
|
||||
}
|
||||
|
||||
int SpeexNB_2_Pcm16( unsigned char* out_buf, unsigned char* in_buf, unsigned int size,
|
||||
unsigned int channels, unsigned int rate, long h_codec )
|
||||
{
|
||||
SpeexState* ss;
|
||||
OneWay *decoder;
|
||||
short* pcm = (short*) out_buf;
|
||||
int frames_out = 0;
|
||||
|
||||
ss = (SpeexState*) h_codec;
|
||||
if (!ss || channels!=1 || rate!=8000)
|
||||
return -1;
|
||||
|
||||
/* Decoder init */
|
||||
if (!(decoder=ss->decoder)) {
|
||||
decoder = ss->decoder = (OneWay*) calloc(1, sizeof(OneWay));
|
||||
if (!decoder)
|
||||
return -1;
|
||||
decoder->state = speex_decoder_init(&speex_nb_mode);
|
||||
speex_decoder_ctl(decoder->state, SPEEX_SET_ENH, &ss->perceptual);
|
||||
speex_bits_init(&decoder->bits);
|
||||
}
|
||||
|
||||
speex_bits_read_from(&decoder->bits, (char*)in_buf, size);
|
||||
|
||||
/* We don't know where frame boundaries are,
|
||||
but the minimum frame size is 43 */
|
||||
while (speex_bits_remaining(&decoder->bits)>40) {
|
||||
int ret;
|
||||
|
||||
#ifdef NOFPU
|
||||
ret = speex_decode_int(decoder->state, &decoder->bits, pcm);
|
||||
pcm+=SPEEX_NB_SAMPLES_PER_FRAME;
|
||||
#else
|
||||
int i;
|
||||
ret = speex_decode(decoder->state, &decoder->bits, decoder->pool);
|
||||
for (i=0; i<SPEEX_NB_SAMPLES_PER_FRAME; i++) {
|
||||
*pcm++ = (short) decoder->pool[i];
|
||||
}
|
||||
#endif
|
||||
|
||||
if (ret==-2) {
|
||||
ERROR("while calling speex_decode\n");
|
||||
return -1;
|
||||
}
|
||||
|
||||
if (ret==-1) break;
|
||||
|
||||
frames_out++;
|
||||
}
|
||||
|
||||
return frames_out*SPEEX_NB_SAMPLES_PER_FRAME*sizeof(short);
|
||||
}
|
||||
Loading…
Reference in new issue