App下载 微信公众号

PCM音频压缩A-Law算法,uLaw

技术 · 移动开发 · Android/ 作者【吾非言】/ 发布于2022-12-19/ 1.08k次浏览
2022 12/19 9:42
摘要: An A-law algorithm is a standard companding algorithm, used in European 8-bit PCM digital communications systems to optimize, i.e. modify, the dynamic range of an analog signal for digitizing. It is one of two versions of the G.711 standard from ITU-T, the other version being the similar μ-law, used in North America and Japan.

参考相关:“关于pcm音频头与a-law音频头的互换代码”
http://blog.csdn.net/fjhyy/article/details/6593049
https://baike.baidu.com/item/A%E5%BE%8B/5923040
http://www.sohu.com/a/147745747_464086
https://en.wikipedia.org/wiki/A-law_algorithm

背景:正在做的一个项目,需要网络传输音频,音频格式为wav,16位采样率,PCM编码。为了降低传输带宽,需要对音频数据进行压缩。

把16bit的音频数据转为8bit。

最简单的方式是均匀量化, >>8 (右移8位),但这样做会使得声音的噪音变大。

最好的做法是使用非均匀量化(如A-Law),其原理是对于小音量的声音,其蕴含的信息量更大,人耳对小音量

更敏感;而大音量部分则影响没那么大。因此使用非均匀量化的方式,对于小音量部分保留更多的数据,大音量部分

则保留更少的数据。具体实现如下:

这里选择A-Law(A律)算法,也可以用uLaw(μ律),两种算法可相互转化。

令量化器过载电压为1,相当于把输入信号进行归一化,那么A律对数压缩定义为:

当0 <= x <= 1/A时,f(x)=(Ax)/(1+lnA)
当1/A <= x <= 1时,f(x)=(1+lnAx)/(1+lnA)

在现行的国际标准中A=87.6,此时信号很小时(即小信号时),从上式可以看到信号被放大了16倍,
这相当于A压缩率与无压缩特性比较,对于小信号的情况,量化间隔比均匀量化时减小了16倍,
因此,量化误差大大降低;而对于大信号的情况例如x=1,量化间隔比均匀量化时增大了5.47倍,
量化误差增大了。这样实际上就实现了“压大补小”的效果。

图形曲线如下(只画出正数部分,对于负数也是同样的道理)(这里用了归一化):

图片描述

但是要在程序中实现该曲线,比较复杂。因此这里使用8段折线来近似表示。

把x轴划分为不均匀的8份,第一点取1/2处,第二点取1/4处,第三点取1/8处……第七点取1/128.

把y轴划分为均匀的8分段。

代码实现如下:

A-Law算法实现PCM压缩.c

#include "stdio.h"

#define SIGN_BIT    (0x80)      /* Sign bit for a A-law byte. */
#define QUANT_MASK  (0xf)       /* Quantization field mask.   */
#define NSEGS       (8)         /* Number of A-law segments.  */
#define SEG_SHIFT   (4)         /* Left shift for segment number. */
#define SEG_MASK    (0x70)      /* Segment field mask. */

static short seg_end[8] = {0xFF, 0x1FF, 0x3FF, 0x7FF,
                           0xFFF, 0x1FFF, 0x3FFF, 0x7FFF};  //分成不均匀的8个分段,算上负数,总共是16个分段

/* copy from CCITT G.711 specifications */
unsigned char _u2a[128] = { /* u- to A-law conversions */
    1,  1,  2,  2,  3,  3,  4,  4,
    5,  5,  6,  6,  7,  7,  8,  8,
    9,  10, 11, 12, 13, 14, 15, 16,
    17, 18, 19, 20, 21, 22, 23, 24,
    25, 27, 29, 31, 33, 34, 35, 36,
    37, 38, 39, 40, 41, 42, 43, 44,
    46, 48, 49, 50, 51, 52, 53, 54,
    55, 56, 57, 58, 59, 60, 61, 62,
    64, 65, 66, 67, 68, 69, 70, 71,
    72, 73, 74, 75, 76, 77, 78, 79,
    81, 82, 83, 84, 85, 86, 87, 88,
    89, 90, 91, 92, 93, 94, 95, 96,
    97, 98, 99, 100,101,102,103,104,
    105,106,107,108,109,110,111,112,
    113,114,115,116,117,118,119,120,
    121,122,123,124,125,126,127,128
};

unsigned char _a2u[128] = { /* A- to u-law conversions */
    1,  3,  5,  7,  9,  11, 13, 15,
    16, 17, 18, 19, 20, 21, 22, 23,
    24, 25, 26, 27, 28, 29, 30, 31,
    32, 32, 33, 33, 34, 34, 35, 35,
    36, 37, 38, 39, 40, 41, 42, 43,
    44, 45, 46, 47, 48, 48, 49, 49,
    50, 51, 52, 53, 54, 55, 56, 57,
    58, 59, 60, 61, 62, 63, 64, 64,
    65, 66, 67, 68, 69, 70, 71, 72,
    73, 74, 75, 76, 77, 78, 79, 79,
    80, 81, 82, 83, 84, 85, 86, 87,
    88, 89, 90, 91, 92, 93, 94, 95,
    96, 97, 98, 99, 100,101,102,103,
    104,105,106,107,108,109,110,111,
    112,113,114,115,116,117,118,119,
    120,121,122,123,124,125,126,127
};

static int search(int val,short *table,int size)
{
    int     i;
    for (i = 0; i < size; i++) {
        if (val <= *table++)
            return (i);
    }
    return (size);
}

/*********************************************************************
 * 输入参数范围 :-32768~32767
 * 返回8位无符号整数
 * linear2alaw() - Convert a 16-bit linear PCM value to 8-bit A-law
 *
 * linear2alaw() accepts an 16-bit integer and encodes it as A-law data.
 *
 *  Linear Input Code       Compressed Code
 *  -----------------       ------------------
 *  0000000wxyza            000wxyz
 *  0000001wxyza            001wxyz
 *  000001wxyzab            010wxyz
 *  00001wxyzabc            011wxyz
 *  0001wxyzabcd            100wxyz
 *  001wxyzabcde            101wxyz
 *  01wxyzabcdef            110wxyz
 *  1wxyzabcdefg            111wxyz
 *
 * For further information see John C. Bellamy's Digital Telephony, 1982,
 * John Wiley & Sons, pps 98-111 and 472-476.
 *********************************************************************/
unsigned char linear2alaw(int pcm_val)  /* 2's complement (16-bit range) */
{
    int             mask;
    int             seg;
    unsigned char   aval;

    if (pcm_val >= 0) {
        mask = 0xD5;        /* sign (7th) bit = 1 */
    } else {
        mask = 0x55;        /* sign bit = 0 */
        //pcm_val = -pcm_val - 8;
        pcm_val = -pcm_val - 1;
    }

    /* Convert the scaled magnitude to segment number. */
    seg = search(pcm_val, seg_end, 8);  //返回pcm_val属于哪个分段

    /* Combine the sign, segment, and quantization bits. */

    if (seg >= 8)        /* out of range, return maximum value. */
        return (0x7F ^ mask);
    else {
        aval = seg << SEG_SHIFT;  //aval为每一段的偏移,分段量化后的数据需要加上该偏移(aval)
        //分段量化
        //量化方法: (pcm_val-分段值),然后取有效的高4位   (0分段例外)
        //比如 pcm_val = 0x7000 ,那么seg=7 ,第7段的范围是0x4000~0x7FFF ,段偏移aval=7<<4=0x7F
        //0x7000-0x4000=0x3000 ,然后取有效的高4位,即右移10(seg+3),0x3000>>10=0xC
        //上一步等效为:(0x7000>>10)&0xF=0xC 。也就是: (pcm_val >> (seg + 3)) & QUANT_MASK
        //然后加上段偏移 0x7F(aval) ,加法等效于或运算,即 |aval

        if (seg < 2)
            aval |= (pcm_val >> 4) & QUANT_MASK;  //0、1段折线的斜率一样
        else
            aval |= (pcm_val >> (seg + 3)) & QUANT_MASK;
        return (aval ^ mask);    //异或0x55,目的是尽量避免出现连续的0,或连续的1,提高传输过程的可靠性
    }
}

/*********************************************************************
 *    alaw2linear() - Convert an A-law value to 16-bit linear PCM
 *********************************************************************/
int alaw2linear(unsigned char a_val)
{
    int     t;
    int     seg;

    a_val ^= 0x55;

    t = (a_val & QUANT_MASK) << 4;
    seg = ((unsigned)a_val & SEG_MASK) >> SEG_SHIFT;
    switch (seg) {
    case 0:
        t += 8;
        break;
    case 1:
        t += 0x108;
        break;
    default:
        t += 0x108;
        t <<= seg - 1;
    }
    return ((a_val & SIGN_BIT) ? t : -t);
}


int main(int argc, char* argv[])
{

    int cmp;
    int dcmp;
    int d;

    while(1)
    {
        scanf("%d",&d);     //输入16位的整数
        printf("input:%d \n",d);
        cmp=linear2alaw(d);     //压缩成8位整数
        dcmp=alaw2linear(cmp);  //解压缩为16位整数

        printf("cmp:%d , dcmp:%d \n",cmp,dcmp);
    }

    return 0;
}

运行结果:

图片描述

对1753压缩后得到238,然后把238解压还原得到1760. 可以看出,还原后的数据,和原始数据还是相当接近的。输入比较小的时候,还原误差小;输入大时,误差大。

阅读原文

感谢您使用伴职平台,如有侵权,请投诉删除!

全部评价

最新
查看更多评论 加载

猜你喜欢

换一批