webrtc（一） agc

一、参数

（一）、mode

enum Mode {
// Adaptive mode intended for use if an analog volume control is available
// on the capture device. It will require the user to provide coupling
// between the OS mixer controls and AGC through the |stream_analog_level()|
// functions.
//
// It consists of an analog gain prescription for the audio device and a
// digital compression stage.
kAdaptiveAnalog,

// Adaptive mode intended for situations in which an analog volume control
// is unavailable. It operates in a similar fashion to the adaptive analog
// mode, but with scaling instead applied in the digital domain. As with
// the analog mode, it additionally uses a digital compression stage.
kAdaptiveDigital,

// Fixed mode which enables only the digital compression stage also used by
// the two adaptive modes.
//
// It is distinguished from the adaptive modes by considering only a
// short time-window of the input signal. It applies a fixed gain through
// most of the input level range, and compresses (gradually reduces gain
// with increasing level) the input signal at higher levels. This mode is
// preferred on embedded devices where the capture signal level is
// predictable, so that a known gain can be applied.
kFixedDigital
};

kAdaptiveAnalog

适合用desktop platforms，模拟信号可控音量

kAdaptiveDigital

适合用mobile device，没有音量控制

kFixedDigital

固定增益

（二）、parameter

以下摘自于 http://rg4.net/archives/797.html

——————————————-

AGC parameters:

——————————————-

1. Target level DBFS

——————————————-

According to the description of webrtc:

Change of this parameter will set the target peak |level| (or envelope) of the AGC in dBFs (decibels from digital full-scale).

The convention is to use positive values.

For instance, passing in a value of 3 corresponds to -3 dBFs, or a target level 3 dB below full-scale.

Value range: limited to [0, 31].

TODO(ajm): use a negative value here instead, if/when VoE will similarly update its interface.

——————————————-

2. Compression gain DB

——————————————-

Sets the maximum |gain| the digital compression stage may apply, in dB. A higher number corresponds to greater compression, while a value of 0 will leave the signal uncompressed.

Value range: limited to [0, 90]

——————————————-

3. Enable limiter or not

——————————————-

When enabled, the compression stage will hard limit the signal to the target level. Otherwise, the signal will be compressed but not limited above the target level.

——————————————-

4. Analog level limits

——————————————-

Sets the |minimum| and |maximum| analog levels of the audio capture device.

Must be set if and only if an analog mode is used.

Value range: limited to [0, 65535].

——————————————-

5. Stream saturated or not

——————————————-

Returns true if the AGC has detected a saturation event (period where the signal reaches digital full-scale) in the current frame and the analog level cannot be reduced.

This could be used as an indicator to reduce or disable analog mic gain at the audio HAL.

以上摘自于 http://rg4.net/archives/797.html

二、使用

由于不同版本的代码差距较大，首先声明下面使用例子来自于下面的commit

commit 99f8566524b3764bf84b7f4cbd8232bdf6c142a5
Author: kjellander <kjellander@webrtc.org>
Date:   Sun Nov 29 20:50:44 2015 -0800

使用自动增益控制放大的音量比较有限，所以使用了固定增益模式

int agcTest2(char *filename, char *outfilename, int mode)
{
//init agc
void *agcInst = WebRtcAgc_Create();
int minLevel = 0;
int maxLevel = 255;
int agcMode  = kAgcModeFixedDigital;
int fs       = 16000;
int status   = 0;
status = WebRtcAgc_Init(agcInst, minLevel, maxLevel, agcMode, fs);
if(status != 0)
{
printf("failed in WebRtcAgc_Init\n");
return -1;
}
WebRtcAgcConfig agcConfig;
agcConfig.compressionGaindB = 20;//在Fixed模式下，越大声音越大
agcConfig.limiterEnable = 1;
agcConfig.targetLevelDbfs = 3;   //dbfs表示相对于full scale的下降值，0表示full scale，越小声音越大
status = WebRtcAgc_set_config(agcInst, agcConfig);
if(status != 0)
{
printf("failed in WebRtcAgc_set_config\n");
return -1;
}

//alloc
FILE *infp=fopen(filename,"r");
int nBands = 1;
int frameSize = 160;//10ms对应于160个short
short **pData = (short**)malloc(nBands*sizeof(short*));
pData[0] = (short*)malloc(frameSize*sizeof(short));
short **pOutData = (short**)malloc(nBands*sizeof(short*));
pOutData[0] = (short*)malloc(frameSize*sizeof(short));

//process
FILE *outfp = fopen(outfilename,"w");
int len = frameSize;
int micLevelIn = 0;
int micLevelOut = 0;
while(len > 0)
{
memset(pData[0], 0, frameSize*sizeof(short));
len = fread(pData[0], sizeof(short), frameSize, infp);

int inMicLevel  = micLevelOut;
int outMicLevel = 0;
uint8_t saturationWarning;
status = WebRtcAgc_Process(agcInst, pData, nBands, frameSize, pOutData, inMicLevel, &outMicLevel, 0, &saturationWarning);
if (status != 0)
{
printf("failed in WebRtcAgc_Process\n");
return -1;
}
if (saturationWarning == 1)
{
printf("saturationWarning\n");
}
micLevelIn = outMicLevel;
//write file
len = fwrite(pOutData[0], sizeof(short), len, outfp);
}
fclose(infp);
fclose(outfp);

WebRtcAgc_Free(agcInst);
free(pData[0]);
free(pData);
free(pOutData[0]);
free(pOutData);

代码使用参考：webrtc/modules/audio_processing/gain_control_impl.cc

三、原理

经过google邮件和论坛考证，确定没有公布出相关的参考文献和原理说明文档。

想不通为什么代码都开源了，好歹说几个参考文献啊！源码也没怎么看，大致看了下注释，所以我在使用过程中，参数的含义和使用基本上靠猜和试验，后续有时间读几篇ns的文献再来细聊，有什么错误麻烦指正！