view amr.c @ 1306:799839d1e2e1 libavcodec

golomb rice codes use gradients instead of prediction errors as context model store independant quantization tables for each point merge contexts with opposit sign

/*
 * AMR Audio decoder stub
 * Copyright (c) 2003 the ffmpeg project
 *
 * This library 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; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This library 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
 */
 /*
    This code implements amr-nb audio encoder/decoder through external reference
    code from www.3gpp.org. The licence of the code from 3gpp is unclear so you
    have to download the code separately. Two versions exists: One fixed-point
    and one with floats. For some reason the float-encoder is significant faster
    atleast on a P4 1.5GHz (0.9s instead of 9.9s on a 30s audio clip at MR102).
    
    The fixed-point (TS26.073) can be downloaded from:
    http://www.3gpp.org/ftp/Specs/latest/Rel-5/26_series/26073-510.zip
    Extract the soure into ffmpeg/libavcodec/amr
    To use the float version run "./configure" with "--enable-amr-nb-fixed"
    
    The float version (default) can be downloaded from:
    http://www.3gpp.org/ftp/Specs/latest/Rel-5/26_series/26104-510.zip
    Extract the soure into ffmpeg/libavcodec/amr_float
    
    The specification for amr-nb can be found in TS 26.071
    (http://www.3gpp.org/ftp/Specs/html-info/26071.htm) and some other
    info at http://www.3gpp.org/ftp/Specs/html-info/26-series.htm
    
    In the future support for AMR-WB might also be included here.
    Reference code exist in TS26.173 and TS 26.204.
 
 */
#define DEBUG
//#define AMR_NB_FIXED
#include "../config.h"
#include "avcodec.h"

#ifdef AMR_NB_FIXED

#define MMS_IO

#include "amr/sp_dec.h"
#include "amr/d_homing.h"
#include "amr/typedef.h"
#include "amr/sp_enc.h"
#include "amr/sid_sync.h"
#include "amr/e_homing.h"

#else
#include "amr_float/interf_dec.h"
#include "amr_float/interf_enc.h"
#endif

/* Common code for fixed and float version*/
typedef struct AMR_bitrates
{
    int startrate;
    int stoprate;
    enum Mode mode;
    
} AMR_bitrates;

/* Match desired bitrate with closest one*/
static enum Mode getBitrateMode(int bitrate)
{
    /* Adjusted so that all bitrates can be used from commandline where
       only a multiple of 1000 can be specified*/
    AMR_bitrates rates[]={ {0,4999,MR475}, //4
                           {5000,5899,MR515},//5
                           {5900,6699,MR59},//6
                           {6700,7000,MR67},//7
                           {7001,7949,MR74},//8
                           {7950,9999,MR795},//9
                           {10000,11999,MR102},//10
                           {12000,64000,MR122},//12
                           
                         };
    int i;
    for(i=0;i<sizeof(rates);i++)
    {
        if(rates[i].startrate<=bitrate && rates[i].stoprate>=bitrate)
        {
            return(rates[i].mode);
        }
    }
    /*Return highest possible*/
    return(MR122);
}

#ifdef AMR_NB_FIXED
/* fixed point version*/
/* frame size in serial bitstream file (frame type + serial stream + flags) */
#define SERIAL_FRAMESIZE (1+MAX_SERIAL_SIZE+5)

typedef struct AMRContext {
    int frameCount;
    Speech_Decode_FrameState *speech_decoder_state;
    enum RXFrameType rx_type;
    enum Mode mode;
    Word16 reset_flag;
    Word16 reset_flag_old;

    enum Mode enc_bitrate;
    Speech_Encode_FrameState *enstate;
    sid_syncState *sidstate;
    enum TXFrameType tx_frametype;
    

} AMRContext;

static int amr_nb_decode_init(AVCodecContext * avctx)
{
    AMRContext *s = avctx->priv_data;
    s->frameCount=0;
    s->speech_decoder_state=NULL;
    s->rx_type = (enum RXFrameType)0;
    s->mode= (enum Mode)0;
    s->reset_flag=0;
    s->reset_flag_old=1;
    
    if(Speech_Decode_Frame_init(&s->speech_decoder_state, "Decoder"))
    {
        printf("Speech_Decode_Frame_init error\n");
        return -1;
    }
    return 0;
}

static int amr_nb_encode_init(AVCodecContext * avctx)
{
    AMRContext *s = avctx->priv_data;
    s->frameCount=0;
    s->speech_decoder_state=NULL;
    s->rx_type = (enum RXFrameType)0;
    s->mode= (enum Mode)0;
    s->reset_flag=0;
    s->reset_flag_old=1;
    
    if(avctx->sample_rate!=8000)
    {
#ifdef DEBUG
        printf("Only 8000Hz sample rate supported\n");
#endif
        return -1;
    }

    if(avctx->channels!=1)
    {
#ifdef DEBUG
        printf("Only mono supported\n");
#endif
        return -1;
    }

    avctx->frame_size=160;
    avctx->coded_frame= avcodec_alloc_frame();

    if(Speech_Encode_Frame_init(&s->enstate, 0, "encoder") || sid_sync_init (&s->sidstate))
    {
#ifdef DEBUG
        printf("Speech_Encode_Frame_init error\n");
#endif
        return -1;
    }

    s->enc_bitrate=getBitrateMode(avctx->bit_rate);

    return 0;
}

static int amr_nb_encode_close(AVCodecContext * avctx)
{
    AMRContext *s = avctx->priv_data;
    Speech_Encode_Frame_exit(&s->enstate);
    sid_sync_exit (&s->sidstate);
    av_freep(&avctx->coded_frame);
    return 0;
}

static int amr_nb_decode_close(AVCodecContext * avctx)
{
    AMRContext *s = avctx->priv_data;
    Speech_Decode_Frame_exit(&s->speech_decoder_state);
    return 0;
}

static int amr_nb_decode_frame(AVCodecContext * avctx,
            void *data, int *data_size,
            uint8_t * buf, int buf_size)
{
    AMRContext *s = avctx->priv_data;

    uint8_t*amrData=buf;
    int offset=0;

    UWord8 toc, q, ft;
    
    Word16 serial[SERIAL_FRAMESIZE];   /* coded bits */
    Word16 *synth;
    UWord8 *packed_bits;
    *data_size=0;

    static Word16 packed_size[16] = {12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0};
    int i;

    //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);

    synth=data;

//    while(offset<buf_size)
    {
        toc=amrData[offset];
        /* read rest of the frame based on ToC byte */
        q  = (toc >> 2) & 0x01;
        ft = (toc >> 3) & 0x0F;

        //printf("offset=%d, packet_size=%d amrData= 0x%X %X %X %X\n",offset,packed_size[ft],amrData[offset],amrData[offset+1],amrData[offset+2],amrData[offset+3]);

        offset++;

        packed_bits=amrData+offset;

        offset+=packed_size[ft];

        //Unsort and unpack bits
        s->rx_type = UnpackBits(q, ft, packed_bits, &s->mode, &serial[1]);

        //We have a new frame
        s->frameCount++;

        if (s->rx_type == RX_NO_DATA) 
        {
            s->mode = s->speech_decoder_state->prev_mode;
        }
        else {
            s->speech_decoder_state->prev_mode = s->mode;
        }
        
        /* if homed: check if this frame is another homing frame */
        if (s->reset_flag_old == 1)
        {
            /* only check until end of first subframe */
            s->reset_flag = decoder_homing_frame_test_first(&serial[1], s->mode);
        }
        /* produce encoder homing frame if homed & input=decoder homing frame */
        if ((s->reset_flag != 0) && (s->reset_flag_old != 0))
        {
            for (i = 0; i < L_FRAME; i++)
            {
                synth[i] = EHF_MASK;
            }
        }
        else
        {     
            /* decode frame */
            Speech_Decode_Frame(s->speech_decoder_state, s->mode, &serial[1], s->rx_type, synth);
        }

        //Each AMR-frame results in 160 16-bit samples
        *data_size+=160*2;
        synth+=160;
        
        /* if not homed: check whether current frame is a homing frame */
        if (s->reset_flag_old == 0)
        {
            /* check whole frame */
            s->reset_flag = decoder_homing_frame_test(&serial[1], s->mode);
        }
        /* reset decoder if current frame is a homing frame */
        if (s->reset_flag != 0)
        {
            Speech_Decode_Frame_reset(s->speech_decoder_state);
        }
        s->reset_flag_old = s->reset_flag;
        
    }
    return offset;
}


static int amr_nb_encode_frame(AVCodecContext *avctx,
			    unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
{
    short serial_data[250] = {0};

    AMRContext *s = avctx->priv_data;
    int written;
   
    s->reset_flag = encoder_homing_frame_test(data);
    
    Speech_Encode_Frame(s->enstate, s->enc_bitrate, data, &serial_data[1], &s->mode); 
    
    /* add frame type and mode */
    sid_sync (s->sidstate, s->mode, &s->tx_frametype);
    
    written = PackBits(s->mode, s->enc_bitrate, s->tx_frametype, &serial_data[1], frame);
    
    if (s->reset_flag != 0)
    {
        Speech_Encode_Frame_reset(s->enstate);
        sid_sync_reset(s->sidstate);
    }
    return written;
}


#else /* Float point version*/

typedef struct AMRContext {
    int frameCount;
    void * decState;
    int *enstate;
    enum Mode enc_bitrate;
} AMRContext;

static int amr_nb_decode_init(AVCodecContext * avctx)
{
    AMRContext *s = avctx->priv_data;
    s->frameCount=0;
    s->decState=Decoder_Interface_init();
    if(!s->decState)
    {
        printf("Decoder_Interface_init error\r\n");
        return -1;
    }
    return 0;
}

static int amr_nb_encode_init(AVCodecContext * avctx)
{
    AMRContext *s = avctx->priv_data;
    s->frameCount=0;
    
    if(avctx->sample_rate!=8000)
    {
#ifdef DEBUG
        printf("Only 8000Hz sample rate supported\n");
#endif
        return -1;
    }

    if(avctx->channels!=1)
    {
#ifdef DEBUG
        printf("Only mono supported\n");
#endif
        return -1;
    }

    avctx->frame_size=160;
    avctx->coded_frame= avcodec_alloc_frame();

    s->enstate=Encoder_Interface_init(0);
    if(!s->enstate)
    {
        printf("Encoder_Interface_init error\n");
        return -1;
    }

    s->enc_bitrate=getBitrateMode(avctx->bit_rate);

    return 0;
}

static int amr_nb_decode_close(AVCodecContext * avctx)
{
    AMRContext *s = avctx->priv_data;
    Decoder_Interface_exit(s->decState);
    return 0;
}

static int amr_nb_encode_close(AVCodecContext * avctx)
{
    AMRContext *s = avctx->priv_data;
    Encoder_Interface_exit(s->enstate);
    av_freep(&avctx->coded_frame);
    return 0;
}

static int amr_nb_decode_frame(AVCodecContext * avctx,
            void *data, int *data_size,
            uint8_t * buf, int buf_size)
{
    AMRContext *s = (AMRContext*)avctx->priv_data;

    uint8_t*amrData=buf;
    int offset=0;
    static short block_size[16]={ 12, 13, 15, 17, 19, 20, 26, 31, 5, 0, 0, 0, 0, 0, 0, 0 };
    enum Mode dec_mode;
    int packet_size;
    *data_size=0;

    //printf("amr_decode_frame data_size=%i buf=0x%X buf_size=%d frameCount=%d!!\n",*data_size,buf,buf_size,s->frameCount);

    while(offset<buf_size)
    {
    	dec_mode = (amrData[offset] >> 3) & 0x000F;
    	packet_size = block_size[dec_mode];
    
        s->frameCount++;
        //printf("offset=%d, packet_size=%d amrData= 0x%X %X %X %X\n",offset,packet_size,amrData[offset],amrData[offset+1],amrData[offset+2],amrData[offset+3]);
    	/* call decoder */
    	Decoder_Interface_Decode(s->decState, &amrData[offset], data+*data_size, 0);
    	*data_size+=160*2;
   
        offset+=packet_size+1; 
    }
    return buf_size;
}

static int amr_nb_encode_frame(AVCodecContext *avctx,
			    unsigned char *frame/*out*/, int buf_size, void *data/*in*/)
{
    AMRContext *s = (AMRContext*)avctx->priv_data;
    int written;

    written = Encoder_Interface_Encode(s->enstate, 
        s->enc_bitrate, 
        data, 
        frame, 
        0);

    return written;
}

#endif

AVCodec amr_nb_decoder =
{
    "amr_nb",
    CODEC_TYPE_AUDIO,
    CODEC_ID_AMR_NB,
    sizeof(AMRContext),
    amr_nb_decode_init,
    NULL,
    amr_nb_decode_close,
    amr_nb_decode_frame,
};

AVCodec amr_nb_encoder =
{
    "amr_nb",
    CODEC_TYPE_AUDIO,
    CODEC_ID_AMR_NB,
    sizeof(AMRContext),
    amr_nb_encode_init,
    amr_nb_encode_frame,
    amr_nb_encode_close,
    NULL,
};