/* FloorDProbs: apply floor to given discrete prob set */
static void FloorDProbs(ShortVec mixes, int M, float floor)
{
float sum,fsum,scale,fltWt;
int m;
sum = fsum = 0.0;
for (m=1; m<=M; m++) {
fltWt = Short2DProb(mixes[m]);
if (fltWt>floor)
sum += fltWt;
else {
fsum += floor;
mixes[m] = DProb2Short(floor);
}
}
if (fsum>1.0) HError(2327,"FloorDProbs: Floor sum too large (%f)",fsum);
if (fsum == 0.0) return;
if (sum == 0.0) HError(2328,"FloorDProbs: No probabilities above floor");
scale = (1.0-fsum)/sum;
for (m=1; m<=M; m++){
fltWt = Short2DProb(mixes[m]);
if (fltWt>floor)
mixes[m] = DProb2Short(fltWt*scale);
}
}
/* UpdateWeights: use acc values to calc new estimate of mix weights */
void UpdateWeights(HLink hmm)
{
int i,s,m,M=0,N;
float x,occi;
WALink wa;
StateElem *se;
StreamElem *ste;
MLink q;
q=FindMacroStruct(&hset,'h',hmm);
N = hmm->numStates;
se = hmm->svec+2;
for (i=2; i<N; i++,se++){
ste = se->info->pdf+1;
for (s=1;s<=nStreams; s++,ste++){
wa = (WALink)ste->hook;
if (wa != NULL) {
switch(hsKind){
case TIEDHS:
M = hset.tmRecs[s].nMix;
break;
case DISCRETEHS:
M = ste->nMix;
break;
}
occi = wa->occ;
if (occi>0) {
for (m=1; m<=M; m++){
x = wa->c[m]/occi;
if (x>1.0){
if (x>1.001)
HError(-2490,"UpdateWeights: Model [%s]: mix too big in %d.%d.%d",q->id->name,i,s,m);
x = 1.0;
}
switch (hsKind){
case TIEDHS:
ste->spdf.tpdf[m] = (x>MINMIX) ? x : 0;
break;
case DISCRETEHS:
ste->spdf.dpdf[m] = (x>MINMIX) ? DProb2Short(x) :DLOGZERO;
break;
}
}
if (mixWeightFloor>0.0){
switch (hsKind){
case DISCRETEHS:
FloorDProbs(ste->spdf.dpdf,M,mixWeightFloor);
break;
case TIEDHS:
FloorTMMixes(ste->spdf.tpdf,M,mixWeightFloor);
break;
}
}
}else
HError(-2427,"UpdateWeights: Model [%s]: no use of mixtures in %d.%d",q->id->name,i,s);
ste->hook = NULL;
}
}
}
}
/* UpDProbs: update given mixture weights */
void UpDProbs(int i, int s, int M, WtAcc *wa, ShortVec dw)
{
int m;
float x,sum=0.0;
if (wa->occ == 0.0)
HError(2127,"UpDProbs: zero occ i=%d/s=%d",i,s);
for (m=1; m<=M; m++){
sum += wa->c[m];
x = wa->c[m] / wa->occ;
if (x<mixWeightFloor) x = mixWeightFloor;
dw[m] = DProb2Short(x);
}
if (fabs(sum-wa->occ)/sum > 0.001)
HError(2190,"UpDProbs: dprob weight sum error");
}
/* UniformSegment: and cluster within each state/segment */
void UniformSegment(void)
{
Sequence **seqMat; /* Matrix [2..numStates-1][1..numStreams]*/
Sequence seq;
int count,size,i,s,n,m,M,j,k;
long int vqidx;
ClusterSet *cset;
Cluster *c;
StreamElem *ste;
Covariance cov;
CovKind ck;
MixPDF *mp;
TMixRec *tmRec = NULL;
ShortVec dw;
float x,z;
Vector floor;
if (trace & T_UNI)
printf(" Uniform Segmentation\n");
seqMat = CreateSeqMat();
cov.var = NULL; /* dummy */
UCollectData(seqMat);
if (trace&T_UNI) ShowSeqMat(seqMat);
/* Cluster Each State/Stream and Init HMM Parms */
for (n=2; n<nStates; n++) {
if (trace&T_UNI) printf(" state %d ",n);
for (s=1; s<=nStreams; s++){
size = hset.swidth[s];
floor = vFloor[s];
ste = hmmLink->svec[n].info->pdf+s;
if (hset.hsKind == TIEDHS){
tmRec = &(hset.tmRecs[s]);
M = tmRec->nMix;
tmRec->topM = tmRec->nMix;
}
else
M = ste->nMix;
if (trace&T_UNI) printf(" stream %d\n",s);
seq = seqMat[n][s];
switch (hset.hsKind){
case PLAINHS:
case SHAREDHS:
ck = ste->spdf.cpdf[1].mpdf->ckind;
cset = FlatCluster(&clustSetStack,seq,M,NULLC,ck,cov);
if (trace&T_UNI) ShowClusterSet(cset);
for (m=1; m<=M; m++){
mp = ste->spdf.cpdf[m].mpdf;
if (mp->ckind != ck)
HError(2123,"UniformSegment: different covkind within a mix\n");
c = cset->cl+m;
if (uFlags&UPMIXES)
ste->spdf.cpdf[m].weight = (float)c->csize/(float)seq->nItems;
if (uFlags&UPMEANS)
CopyVector(c->vCtr,mp->mean);
if (uFlags&UPVARS)
switch(ck){
case DIAGC:
for (j=1; j<=size; j++){
z= c->cov.var[j];
mp->cov.var[j] = (z<floor[j])?floor[j]:z;
}
break;
case FULLC:
for (j=1; j<=size; j++){
for (k=1; k<j; k++) {
mp->cov.inv[j][k] = c->cov.inv[j][k];
}
z = c->cov.inv[j][j];
mp->cov.inv[j][j] = (z<floor[j])?floor[j]:z;
}
break;
default:
HError(2124,"UniformSegment: bad cov kind %d\n",ck);
}
}
break;
case DISCRETEHS:
count = 0; dw = ste->spdf.dpdf;
ZeroShortVec(dw);
for (i=1; i<=seq->nItems; i++){
vqidx = (long int)GetItem(seq,i);
if (vqidx<1 || vqidx>M)
HError(2170,"UniformSegment: vqidx out of range[%ld]",vqidx);
++dw[vqidx]; ++count;
}
for (m=1; m<=M; m++){
x = (float)dw[m]/(float)count;
if (x<mixWeightFloor) x = mixWeightFloor;
dw[m] = DProb2Short(x);
}
break;
case TIEDHS:
ck = tmRec->mixes[1]->ckind;
cset = FlatCluster(&clustSetStack,seq,M,NULLC,ck,cov);
if (trace&T_UNI) ShowClusterSet(cset);
for (m=1; m<=M; m++){
mp = tmRec->mixes[m];
if (mp->ckind != ck)
HError(2123,"UniformSegment: different covkind within a mix\n");
c = cset->cl+m;
if (uFlags&UPMIXES)
ste->spdf.tpdf[m] = (float)c->csize/(float)seq->nItems;
if (uFlags&UPMEANS)
CopyVector(c->vCtr,mp->mean);
if (uFlags&UPVARS)
switch(ck){
case DIAGC:
for (j=1; j<=size; j++){
z= c->cov.var[j];
mp->cov.var[j] = (z<floor[j])?floor[j]:z;
}
break;
case FULLC:
for (j=1; j<=size; j++){
for (k=1; k<j; k++) {
mp->cov.inv[j][k] = c->cov.inv[j][k];
}
z = c->cov.inv[j][j];
mp->cov.inv[j][j] = (z<floor[j])?floor[j]:z;
}
break;
default:
HError(2124,"UniformSegment: bad cov kind %d\n",ck);
}
}
break;
}
ResetHeap(&clustSetStack);
}
if ((hset.hsKind == PLAINHS) || (hset.hsKind == SHAREDHS))
FixGConsts(hmmLink);
}
ResetHeap(&sequenceStack);
}
/* UpdateWeights: use acc values to calc new estimate of mix weights */
static void UpdateWeights(HMMSet *hset, int px, HLink hmm)
{
int i,s,m,M=0,N,vSize;
float x,occi,denom,tmp;
WtAcc *wa;
StateElem *se;
StreamElem *ste;
MixtureElem *me;
N = hmm->numStates;
se = hmm->svec+2;
for (i=2; i<N; i++,se++){
ste = se->info->pdf+1;
for (s=1;s<=S; s++,ste++){
wa = (WtAcc *)ste->info->hook;
switch (hset->hsKind){
case TIEDHS:
M=hset->tmRecs[s].nMix;
break;
case DISCRETEHS:
case PLAINHS:
case SHAREDHS:
M=ste->info->nMix;
break;
}
if (wa != NULL) {
occi = wa->occ;
if (occi>0) {
me = ste->info->spdf.cpdf + 1; denom=0;
for (m=1; m<=M; m++,me++){
vSize = VectorSize(me->mpdf->mean);
tmp = me->weight*vSize*mapTau -1;
if (tmp<0) tmp = 0;
denom += tmp;
}
me = ste->info->spdf.cpdf + 1;
for (m=1; m<=M; m++,me++){
vSize = VectorSize(me->mpdf->mean);
tmp = me->weight*vSize*mapTau -1;
if (tmp<0) tmp = 0;
x = (tmp + wa->c[m])/(denom + occi);
if (x>1.0){
if (x>1.001)
HError(2393,"UpdateWeights: Model %d[%s]: mix too big in %d.%d.%d %5.5f",
px,HMMPhysName(hset,hmm),i,s,m,x);
x = 1.0;
}
switch (hset->hsKind){
case TIEDHS:
ste->info->spdf.tpdf[m] = x;
break;
case DISCRETEHS:
ste->info->spdf.dpdf[m]=DProb2Short(x);
break;
case PLAINHS:
case SHAREDHS:
me=ste->info->spdf.cpdf+m;
me->weight = x;
break;
}
}
if (mixWeightFloor>0.0){
FloorMixtures(hset->hsKind,ste->info,M,mixWeightFloor);
}
/* Force a normalisation becomes of weird zeroing .... */
if ((hset->hsKind == PLAINHS) || (hset->hsKind == SHAREDHS)) {
me = ste->info->spdf.cpdf + 1; x=0;
for (m=1; m<=M; m++,me++)
x += me->weight;
if (x>1.001)
HError(-1,"Updating Weights, sum too large (%f)\n",x);
me = ste->info->spdf.cpdf + 1;
for (m=1; m<=M; m++,me++)
me->weight /= x;
}
}
ste->info->hook = NULL;
}
}
}
}
case 'l':
maxSpUtt = GetChkedInt(0,0100000,s);
break;
case 'E':
if (NextArg()!=STRINGARG)
HError(2319,"HERest: parent transform directory expected");
xfInfo.usePaXForm = TRUE;
xfInfo.paXFormDir = GetStrArg();
if (NextArg()==STRINGARG)
xfInfo.paXFormExt = GetStrArg();
if (NextArg() != SWITCHARG)
HError(2319,"HERest: cannot have -E as the last option");
break;
case 'J':
if (NextArg()!=STRINGARG)
HError(2319,"HERest: input transform directory expected");
AddInXFormDir(&hset,GetStrArg());
if (NextArg()==STRINGARG) {
if (xfInfo.inXFormExt == NULL)
xfInfo.inXFormExt = GetStrArg();
else
HError(2319,"HERest: only one input transform extension may be specified");
}
if (NextArg() != SWITCHARG)
HError(2319,"HERest: cannot have -J as the last option");
break;
case 'K':
if (NextArg()!=STRINGARG)
HError(2319,"HERest: output transform directory expected");
xfInfo.outXFormDir = GetStrArg();
if (NextArg()==STRINGARG)
xfInfo.outXFormExt = GetStrArg();
if (NextArg() != SWITCHARG)
HError(2319,"HERest: cannot have -K as the last option");
break;
case 'z':
if (NextArg() != STRINGARG)
HError(2319,"HERest: output TMF file expected");
