7 May
2009
7 May
'09
3:55 p.m.
Michael B. Brutman wrote:
I understand the
principle behind CVSD encoding and decoding, but not enough to start
writing code. I want to write some code takes CVSD encoded data and
displays the waveform on a screen, for a sound editor.
CVSD decoding is a simple algorithm, here's a version which works one bit at a time:
typedef unsigned short ushort;
typedef unsigned short bit; // should be a range 0..1.
#define kCvsdMaxGain 8
#define kCvsdIntegrator 15
ushort gCvsdDecGain=1, gCvsdDecRefSample=32768;
ushort gCvsdDecRecentBits=0xaaaa;
ushort CvsdDec(bit aBit)
{
gCvsdDecRecentBits=(gCvsdDecRecentBits<<1)|aBit;
gCvsdDecRefSample+= (aBit==1)? gCvsdDecGain:-gCvsdDecGain;
if(((gCvsdDecRecentBits&kCvsdIntegrator)==kCvsdIntegrator) ||
((gCvsdDecRecentBits&7)==0) ) {
if(gCvsdDecGain<kCvsdMaxGain)
gCvsdDecGain<<=1;
}
else
kCvsdMaxGain=(kCvsdMaxGain+1)>>1;
return gCvsdDecRefSample;
}
All you have to do is pick the right maximum gain and integrator. Somewhat wrong values
will still produce something recognisable.
-cheers from julz @P