split client and server 1
This commit is contained in:
parent
26a15a0b15
commit
2fa521c3d1
661 changed files with 3058 additions and 2826 deletions
443
java/src/client/audio/jorbis/StaticCodeBook.java
Normal file
443
java/src/client/audio/jorbis/StaticCodeBook.java
Normal file
|
@ -0,0 +1,443 @@
|
|||
/* -*-mode:java; c-basic-offset:2; indent-tabs-mode:nil -*- */
|
||||
/* JOrbis
|
||||
* Copyright (C) 2000 ymnk, JCraft,Inc.
|
||||
*
|
||||
* Written by: 2000 ymnk<ymnk@jcraft.com>
|
||||
*
|
||||
* Many thanks to
|
||||
* Monty <monty@xiph.org> and
|
||||
* The XIPHOPHORUS Company http://www.xiph.org/ .
|
||||
* JOrbis has been based on their awesome works, Vorbis codec.
|
||||
*
|
||||
* This program is free software; you can redistribute it and/or
|
||||
* modify it under the terms of the GNU Library General Public License
|
||||
* as published by the Free Software Foundation; either version 2 of
|
||||
* the License, or (at your option) any later version.
|
||||
|
||||
* This program 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 Library General Public License for more details.
|
||||
*
|
||||
* You should have received a copy of the GNU Library General Public
|
||||
* License along with this program; if not, write to the Free Software
|
||||
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
||||
*/
|
||||
|
||||
package client.audio.jorbis;
|
||||
|
||||
import client.audio.jogg.Buffer;
|
||||
|
||||
class StaticCodeBook{
|
||||
int dim; // codebook dimensions (elements per vector)
|
||||
int entries; // codebook entries
|
||||
int[] lengthlist; // codeword lengths in bits
|
||||
|
||||
// mapping
|
||||
int maptype; // 0=none
|
||||
// 1=implicitly populated values from map column
|
||||
// 2=listed arbitrary values
|
||||
|
||||
// The below does a linear, single monotonic sequence mapping.
|
||||
int q_min; // packed 32 bit float; quant value 0 maps to minval
|
||||
int q_delta; // packed 32 bit float; val 1 - val 0 == delta
|
||||
int q_quant; // bits: 0 < quant <= 16
|
||||
int q_sequencep; // bitflag
|
||||
|
||||
// additional information for log (dB) mapping; the linear mapping
|
||||
// is assumed to actually be values in dB. encodebias is used to
|
||||
// assign an error weight to 0 dB. We have two additional flags:
|
||||
// zeroflag indicates if entry zero is to represent -Inf dB; negflag
|
||||
// indicates if we're to represent negative linear values in a
|
||||
// mirror of the positive mapping.
|
||||
|
||||
int[] quantlist; // map == 1: (int)(entries/dim) element column map
|
||||
// map == 2: list of dim*entries quantized entry vals
|
||||
|
||||
StaticCodeBook(){
|
||||
}
|
||||
|
||||
int pack(Buffer opb){
|
||||
int i;
|
||||
boolean ordered=false;
|
||||
|
||||
opb.write(0x564342, 24);
|
||||
opb.write(dim, 16);
|
||||
opb.write(entries, 24);
|
||||
|
||||
// pack the codewords. There are two packings; length ordered and
|
||||
// length random. Decide between the two now.
|
||||
|
||||
for(i=1; i<entries; i++){
|
||||
if(lengthlist[i]<lengthlist[i-1])
|
||||
break;
|
||||
}
|
||||
if(i==entries)
|
||||
ordered=true;
|
||||
|
||||
if(ordered){
|
||||
// length ordered. We only need to say how many codewords of
|
||||
// each length. The actual codewords are generated
|
||||
// deterministically
|
||||
|
||||
int count=0;
|
||||
opb.write(1, 1); // ordered
|
||||
opb.write(lengthlist[0]-1, 5); // 1 to 32
|
||||
|
||||
for(i=1; i<entries; i++){
|
||||
int _this=lengthlist[i];
|
||||
int _last=lengthlist[i-1];
|
||||
if(_this>_last){
|
||||
for(int j=_last; j<_this; j++){
|
||||
opb.write(i-count, Util.ilog(entries-count));
|
||||
count=i;
|
||||
}
|
||||
}
|
||||
}
|
||||
opb.write(i-count, Util.ilog(entries-count));
|
||||
}
|
||||
else{
|
||||
// length random. Again, we don't code the codeword itself, just
|
||||
// the length. This time, though, we have to encode each length
|
||||
opb.write(0, 1); // unordered
|
||||
|
||||
// algortihmic mapping has use for 'unused entries', which we tag
|
||||
// here. The algorithmic mapping happens as usual, but the unused
|
||||
// entry has no codeword.
|
||||
for(i=0; i<entries; i++){
|
||||
if(lengthlist[i]==0)
|
||||
break;
|
||||
}
|
||||
|
||||
if(i==entries){
|
||||
opb.write(0, 1); // no unused entries
|
||||
for(i=0; i<entries; i++){
|
||||
opb.write(lengthlist[i]-1, 5);
|
||||
}
|
||||
}
|
||||
else{
|
||||
opb.write(1, 1); // we have unused entries; thus we tag
|
||||
for(i=0; i<entries; i++){
|
||||
if(lengthlist[i]==0){
|
||||
opb.write(0, 1);
|
||||
}
|
||||
else{
|
||||
opb.write(1, 1);
|
||||
opb.write(lengthlist[i]-1, 5);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// is the entry number the desired return value, or do we have a
|
||||
// mapping? If we have a mapping, what type?
|
||||
opb.write(maptype, 4);
|
||||
switch(maptype){
|
||||
case 0:
|
||||
// no mapping
|
||||
break;
|
||||
case 1:
|
||||
case 2:
|
||||
// implicitly populated value mapping
|
||||
// explicitly populated value mapping
|
||||
if(quantlist==null){
|
||||
// no quantlist? error
|
||||
return (-1);
|
||||
}
|
||||
|
||||
// values that define the dequantization
|
||||
opb.write(q_min, 32);
|
||||
opb.write(q_delta, 32);
|
||||
opb.write(q_quant-1, 4);
|
||||
opb.write(q_sequencep, 1);
|
||||
|
||||
{
|
||||
int quantvals=0;
|
||||
switch(maptype){
|
||||
case 1:
|
||||
// a single column of (c->entries/c->dim) quantized values for
|
||||
// building a full value list algorithmically (square lattice)
|
||||
quantvals=maptype1_quantvals();
|
||||
break;
|
||||
case 2:
|
||||
// every value (c->entries*c->dim total) specified explicitly
|
||||
quantvals=entries*dim;
|
||||
break;
|
||||
}
|
||||
|
||||
// quantized values
|
||||
for(i=0; i<quantvals; i++){
|
||||
opb.write(Math.abs(quantlist[i]), q_quant);
|
||||
}
|
||||
}
|
||||
break;
|
||||
default:
|
||||
// error case; we don't have any other map types now
|
||||
return (-1);
|
||||
}
|
||||
return (0);
|
||||
}
|
||||
|
||||
// unpacks a codebook from the packet buffer into the codebook struct,
|
||||
// readies the codebook auxiliary structures for decode
|
||||
int unpack(Buffer opb){
|
||||
int i;
|
||||
//memset(s,0,sizeof(static_codebook));
|
||||
|
||||
// make sure alignment is correct
|
||||
if(opb.read(24)!=0x564342){
|
||||
// goto _eofout;
|
||||
clear();
|
||||
return (-1);
|
||||
}
|
||||
|
||||
// first the basic parameters
|
||||
dim=opb.read(16);
|
||||
entries=opb.read(24);
|
||||
if(entries==-1){
|
||||
// goto _eofout;
|
||||
clear();
|
||||
return (-1);
|
||||
}
|
||||
|
||||
// codeword ordering.... length ordered or unordered?
|
||||
switch(opb.read(1)){
|
||||
case 0:
|
||||
// unordered
|
||||
lengthlist=new int[entries];
|
||||
|
||||
// allocated but unused entries?
|
||||
if(opb.read(1)!=0){
|
||||
// yes, unused entries
|
||||
|
||||
for(i=0; i<entries; i++){
|
||||
if(opb.read(1)!=0){
|
||||
int num=opb.read(5);
|
||||
if(num==-1){
|
||||
// goto _eofout;
|
||||
clear();
|
||||
return (-1);
|
||||
}
|
||||
lengthlist[i]=num+1;
|
||||
}
|
||||
else{
|
||||
lengthlist[i]=0;
|
||||
}
|
||||
}
|
||||
}
|
||||
else{
|
||||
// all entries used; no tagging
|
||||
for(i=0; i<entries; i++){
|
||||
int num=opb.read(5);
|
||||
if(num==-1){
|
||||
// goto _eofout;
|
||||
clear();
|
||||
return (-1);
|
||||
}
|
||||
lengthlist[i]=num+1;
|
||||
}
|
||||
}
|
||||
break;
|
||||
case 1:
|
||||
// ordered
|
||||
{
|
||||
int length=opb.read(5)+1;
|
||||
lengthlist=new int[entries];
|
||||
|
||||
for(i=0; i<entries;){
|
||||
int num=opb.read(Util.ilog(entries-i));
|
||||
if(num==-1){
|
||||
// goto _eofout;
|
||||
clear();
|
||||
return (-1);
|
||||
}
|
||||
for(int j=0; j<num; j++, i++){
|
||||
lengthlist[i]=length;
|
||||
}
|
||||
length++;
|
||||
}
|
||||
}
|
||||
break;
|
||||
default:
|
||||
// EOF
|
||||
return (-1);
|
||||
}
|
||||
|
||||
// Do we have a mapping to unpack?
|
||||
switch((maptype=opb.read(4))){
|
||||
case 0:
|
||||
// no mapping
|
||||
break;
|
||||
case 1:
|
||||
case 2:
|
||||
// implicitly populated value mapping
|
||||
// explicitly populated value mapping
|
||||
q_min=opb.read(32);
|
||||
q_delta=opb.read(32);
|
||||
q_quant=opb.read(4)+1;
|
||||
q_sequencep=opb.read(1);
|
||||
|
||||
{
|
||||
int quantvals=0;
|
||||
switch(maptype){
|
||||
case 1:
|
||||
quantvals=maptype1_quantvals();
|
||||
break;
|
||||
case 2:
|
||||
quantvals=entries*dim;
|
||||
break;
|
||||
}
|
||||
|
||||
// quantized values
|
||||
quantlist=new int[quantvals];
|
||||
for(i=0; i<quantvals; i++){
|
||||
quantlist[i]=opb.read(q_quant);
|
||||
}
|
||||
if(quantlist[quantvals-1]==-1){
|
||||
// goto _eofout;
|
||||
clear();
|
||||
return (-1);
|
||||
}
|
||||
}
|
||||
break;
|
||||
default:
|
||||
// goto _eofout;
|
||||
clear();
|
||||
return (-1);
|
||||
}
|
||||
// all set
|
||||
return (0);
|
||||
// _errout:
|
||||
// _eofout:
|
||||
// vorbis_staticbook_clear(s);
|
||||
// return(-1);
|
||||
}
|
||||
|
||||
// there might be a straightforward one-line way to do the below
|
||||
// that's portable and totally safe against roundoff, but I haven't
|
||||
// thought of it. Therefore, we opt on the side of caution
|
||||
private int maptype1_quantvals(){
|
||||
int vals=(int)(Math.floor(Math.pow(entries, 1./dim)));
|
||||
|
||||
// the above *should* be reliable, but we'll not assume that FP is
|
||||
// ever reliable when bitstream sync is at stake; verify via integer
|
||||
// means that vals really is the greatest value of dim for which
|
||||
// vals^b->bim <= b->entries
|
||||
// treat the above as an initial guess
|
||||
while(true){
|
||||
int acc=1;
|
||||
int acc1=1;
|
||||
for(int i=0; i<dim; i++){
|
||||
acc*=vals;
|
||||
acc1*=vals+1;
|
||||
}
|
||||
if(acc<=entries&&acc1>entries){
|
||||
return (vals);
|
||||
}
|
||||
else{
|
||||
if(acc>entries){
|
||||
vals--;
|
||||
}
|
||||
else{
|
||||
vals++;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
void clear(){
|
||||
}
|
||||
|
||||
// unpack the quantized list of values for encode/decode
|
||||
// we need to deal with two map types: in map type 1, the values are
|
||||
// generated algorithmically (each column of the vector counts through
|
||||
// the values in the quant vector). in map type 2, all the values came
|
||||
// in in an explicit list. Both value lists must be unpacked
|
||||
float[] unquantize(){
|
||||
|
||||
if(maptype==1||maptype==2){
|
||||
int quantvals;
|
||||
float mindel=float32_unpack(q_min);
|
||||
float delta=float32_unpack(q_delta);
|
||||
float[] r=new float[entries*dim];
|
||||
|
||||
// maptype 1 and 2 both use a quantized value vector, but
|
||||
// different sizes
|
||||
switch(maptype){
|
||||
case 1:
|
||||
// most of the time, entries%dimensions == 0, but we need to be
|
||||
// well defined. We define that the possible vales at each
|
||||
// scalar is values == entries/dim. If entries%dim != 0, we'll
|
||||
// have 'too few' values (values*dim<entries), which means that
|
||||
// we'll have 'left over' entries; left over entries use zeroed
|
||||
// values (and are wasted). So don't generate codebooks like that
|
||||
quantvals=maptype1_quantvals();
|
||||
for(int j=0; j<entries; j++){
|
||||
float last=0.f;
|
||||
int indexdiv=1;
|
||||
for(int k=0; k<dim; k++){
|
||||
int index=(j/indexdiv)%quantvals;
|
||||
float val=quantlist[index];
|
||||
val=Math.abs(val)*delta+mindel+last;
|
||||
if(q_sequencep!=0)
|
||||
last=val;
|
||||
r[j*dim+k]=val;
|
||||
indexdiv*=quantvals;
|
||||
}
|
||||
}
|
||||
break;
|
||||
case 2:
|
||||
for(int j=0; j<entries; j++){
|
||||
float last=0.f;
|
||||
for(int k=0; k<dim; k++){
|
||||
float val=quantlist[j*dim+k];
|
||||
//if((j*dim+k)==0){System.err.println(" | 0 -> "+val+" | ");}
|
||||
val=Math.abs(val)*delta+mindel+last;
|
||||
if(q_sequencep!=0)
|
||||
last=val;
|
||||
r[j*dim+k]=val;
|
||||
//if((j*dim+k)==0){System.err.println(" $ r[0] -> "+r[0]+" | ");}
|
||||
}
|
||||
}
|
||||
//System.err.println("\nr[0]="+r[0]);
|
||||
}
|
||||
return (r);
|
||||
}
|
||||
return (null);
|
||||
}
|
||||
|
||||
// 32 bit float (not IEEE; nonnormalized mantissa +
|
||||
// biased exponent) : neeeeeee eeemmmmm mmmmmmmm mmmmmmmm
|
||||
// Why not IEEE? It's just not that important here.
|
||||
|
||||
static final int VQ_FEXP=10;
|
||||
static final int VQ_FMAN=21;
|
||||
static final int VQ_FEXP_BIAS=768; // bias toward values smaller than 1.
|
||||
|
||||
// doesn't currently guard under/overflow
|
||||
static long float32_pack(float val){
|
||||
int sign=0;
|
||||
int exp;
|
||||
int mant;
|
||||
if(val<0){
|
||||
sign=0x80000000;
|
||||
val=-val;
|
||||
}
|
||||
exp=(int)Math.floor(Math.log(val)/Math.log(2));
|
||||
mant=(int)Math.rint(Math.pow(val, (VQ_FMAN-1)-exp));
|
||||
exp=(exp+VQ_FEXP_BIAS)<<VQ_FMAN;
|
||||
return (sign|exp|mant);
|
||||
}
|
||||
|
||||
static float float32_unpack(int val){
|
||||
float mant=val&0x1fffff;
|
||||
float exp=(val&0x7fe00000)>>>VQ_FMAN;
|
||||
if((val&0x80000000)!=0)
|
||||
mant=-mant;
|
||||
return (ldexp(mant, ((int)exp)-(VQ_FMAN-1)-VQ_FEXP_BIAS));
|
||||
}
|
||||
|
||||
static float ldexp(float foo, int e){
|
||||
return (float)(foo*Math.pow(2, e));
|
||||
}
|
||||
}
|
Loading…
Add table
Add a link
Reference in a new issue