comparison src/libvorbis-1.3.3/lib/vorbisenc.c @ 1:05aa0afa9217

Bring in flac, ogg, vorbis

+/********************************************************************
+ *                                                                  *
+ * THIS FILE IS PART OF THE OggVorbis SOFTWARE CODEC SOURCE CODE.   *
+ * USE, DISTRIBUTION AND REPRODUCTION OF THIS LIBRARY SOURCE IS     *
+ * GOVERNED BY A BSD-STYLE SOURCE LICENSE INCLUDED WITH THIS SOURCE *
+ * IN 'COPYING'. PLEASE READ THESE TERMS BEFORE DISTRIBUTING.       *
+ *                                                                  *
+ * THE OggVorbis SOURCE CODE IS (C) COPYRIGHT 1994-2009             *
+ * by the Xiph.Org Foundation http://www.xiph.org/                  *
+ *                                                                  *
+ ********************************************************************
+
+ function: simple programmatic interface for encoder mode setup
+ last mod: $Id: vorbisenc.c 17028 2010-03-25 05:22:15Z xiphmont $
+
+ ********************************************************************/
+
+#include <stdlib.h>
+#include <string.h>
+#include <math.h>
+
+#include "vorbis/codec.h"
+#include "vorbis/vorbisenc.h"
+
+#include "codec_internal.h"
+
+#include "os.h"
+#include "misc.h"
+
+/* careful with this; it's using static array sizing to make managing
+   all the modes a little less annoying.  If we use a residue backend
+   with > 12 partition types, or a different division of iteration,
+   this needs to be updated. */
+typedef struct {
+  const static_codebook *books[12][4];
+} static_bookblock;
+
+typedef struct {
+  int res_type;
+  int limit_type; /* 0 lowpass limited, 1 point stereo limited */
+  int grouping;
+  const vorbis_info_residue0 *res;
+  const static_codebook  *book_aux;
+  const static_codebook  *book_aux_managed;
+  const static_bookblock *books_base;
+  const static_bookblock *books_base_managed;
+} vorbis_residue_template;
+
+typedef struct {
+  const vorbis_info_mapping0    *map;
+  const vorbis_residue_template *res;
+} vorbis_mapping_template;
+
+typedef struct vp_adjblock{
+  int block[P_BANDS];
+} vp_adjblock;
+
+typedef struct {
+  int data[NOISE_COMPAND_LEVELS];
+} compandblock;
+
+/* high level configuration information for setting things up
+   step-by-step with the detailed vorbis_encode_ctl interface.
+   There's a fair amount of redundancy such that interactive setup
+   does not directly deal with any vorbis_info or codec_setup_info
+   initialization; it's all stored (until full init) in this highlevel
+   setup, then flushed out to the real codec setup structs later. */
+
+typedef struct {
+  int att[P_NOISECURVES];
+  float boost;
+  float decay;
+} att3;
+typedef struct { int data[P_NOISECURVES]; } adj3;
+
+typedef struct {
+  int   pre[PACKETBLOBS];
+  int   post[PACKETBLOBS];
+  float kHz[PACKETBLOBS];
+  float lowpasskHz[PACKETBLOBS];
+} adj_stereo;
+
+typedef struct {
+  int lo;
+  int hi;
+  int fixed;
+} noiseguard;
+typedef struct {
+  int data[P_NOISECURVES][17];
+} noise3;
+
+typedef struct {
+  int      mappings;
+  const double  *rate_mapping;
+  const double  *quality_mapping;
+  int      coupling_restriction;
+  long     samplerate_min_restriction;
+  long     samplerate_max_restriction;
+
+
+  const int     *blocksize_short;
+  const int     *blocksize_long;
+
+  const att3    *psy_tone_masteratt;
+  const int     *psy_tone_0dB;
+  const int     *psy_tone_dBsuppress;
+
+  const vp_adjblock *psy_tone_adj_impulse;
+  const vp_adjblock *psy_tone_adj_long;
+  const vp_adjblock *psy_tone_adj_other;
+
+  const noiseguard  *psy_noiseguards;
+  const noise3      *psy_noise_bias_impulse;
+  const noise3      *psy_noise_bias_padding;
+  const noise3      *psy_noise_bias_trans;
+  const noise3      *psy_noise_bias_long;
+  const int         *psy_noise_dBsuppress;
+
+  const compandblock  *psy_noise_compand;
+  const double        *psy_noise_compand_short_mapping;
+  const double        *psy_noise_compand_long_mapping;
+
+  const int      *psy_noise_normal_start[2];
+  const int      *psy_noise_normal_partition[2];
+  const double   *psy_noise_normal_thresh;
+
+  const int      *psy_ath_float;
+  const int      *psy_ath_abs;
+
+  const double   *psy_lowpass;
+
+  const vorbis_info_psy_global *global_params;
+  const double     *global_mapping;
+  const adj_stereo *stereo_modes;
+
+  const static_codebook *const *const *const floor_books;
+  const vorbis_info_floor1 *floor_params;
+  const int floor_mappings;
+  const int **floor_mapping_list;
+
+  const vorbis_mapping_template *maps;
+} ve_setup_data_template;
+
+/* a few static coder conventions */
+static const vorbis_info_mode _mode_template[2]={
+  {0,0,0,0},
+  {1,0,0,1}
+};
+
+static const vorbis_info_mapping0 _map_nominal[2]={
+  {1, {0,0}, {0}, {0}, 1,{0},{1}},
+  {1, {0,0}, {1}, {1}, 1,{0},{1}}
+};
+
+#include "modes/setup_44.h"
+#include "modes/setup_44u.h"
+#include "modes/setup_44p51.h"
+#include "modes/setup_32.h"
+#include "modes/setup_8.h"
+#include "modes/setup_11.h"
+#include "modes/setup_16.h"
+#include "modes/setup_22.h"
+#include "modes/setup_X.h"
+
+static const ve_setup_data_template *const setup_list[]={
+  &ve_setup_44_stereo,
+  &ve_setup_44_51,
+  &ve_setup_44_uncoupled,
+
+  &ve_setup_32_stereo,
+  &ve_setup_32_uncoupled,
+
+  &ve_setup_22_stereo,
+  &ve_setup_22_uncoupled,
+  &ve_setup_16_stereo,
+  &ve_setup_16_uncoupled,
+
+  &ve_setup_11_stereo,
+  &ve_setup_11_uncoupled,
+  &ve_setup_8_stereo,
+  &ve_setup_8_uncoupled,
+
+  &ve_setup_X_stereo,
+  &ve_setup_X_uncoupled,
+  &ve_setup_XX_stereo,
+  &ve_setup_XX_uncoupled,
+  0
+};
+
+static void vorbis_encode_floor_setup(vorbis_info *vi,int s,
+                                     const static_codebook *const *const *const books,
+                                     const vorbis_info_floor1 *in,
+                                     const int *x){
+  int i,k,is=s;
+  vorbis_info_floor1 *f=_ogg_calloc(1,sizeof(*f));
+  codec_setup_info *ci=vi->codec_setup;
+
+  memcpy(f,in+x[is],sizeof(*f));
+
+  /* books */
+  {
+    int partitions=f->partitions;
+    int maxclass=-1;
+    int maxbook=-1;
+    for(i=0;i<partitions;i++)
+      if(f->partitionclass[i]>maxclass)maxclass=f->partitionclass[i];
+    for(i=0;i<=maxclass;i++){
+      if(f->class_book[i]>maxbook)maxbook=f->class_book[i];
+      f->class_book[i]+=ci->books;
+      for(k=0;k<(1<<f->class_subs[i]);k++){
+        if(f->class_subbook[i][k]>maxbook)maxbook=f->class_subbook[i][k];
+        if(f->class_subbook[i][k]>=0)f->class_subbook[i][k]+=ci->books;
+      }
+    }
+
+    for(i=0;i<=maxbook;i++)
+      ci->book_param[ci->books++]=(static_codebook *)books[x[is]][i];
+  }
+
+  /* for now, we're only using floor 1 */
+  ci->floor_type[ci->floors]=1;
+  ci->floor_param[ci->floors]=f;
+  ci->floors++;
+
+  return;
+}
+
+static void vorbis_encode_global_psych_setup(vorbis_info *vi,double s,
+                                            const vorbis_info_psy_global *in,
+                                            const double *x){
+  int i,is=s;
+  double ds=s-is;
+  codec_setup_info *ci=vi->codec_setup;
+  vorbis_info_psy_global *g=&ci->psy_g_param;
+
+  memcpy(g,in+(int)x[is],sizeof(*g));
+
+  ds=x[is]*(1.-ds)+x[is+1]*ds;
+  is=(int)ds;
+  ds-=is;
+  if(ds==0 && is>0){
+    is--;
+    ds=1.;
+  }
+
+  /* interpolate the trigger threshholds */
+  for(i=0;i<4;i++){
+    g->preecho_thresh[i]=in[is].preecho_thresh[i]*(1.-ds)+in[is+1].preecho_thresh[i]*ds;
+    g->postecho_thresh[i]=in[is].postecho_thresh[i]*(1.-ds)+in[is+1].postecho_thresh[i]*ds;
+  }
+  g->ampmax_att_per_sec=ci->hi.amplitude_track_dBpersec;
+  return;
+}
+
+static void vorbis_encode_global_stereo(vorbis_info *vi,
+                                        const highlevel_encode_setup *const hi,
+                                        const adj_stereo *p){
+  float s=hi->stereo_point_setting;
+  int i,is=s;
+  double ds=s-is;
+  codec_setup_info *ci=vi->codec_setup;
+  vorbis_info_psy_global *g=&ci->psy_g_param;
+
+  if(p){
+    memcpy(g->coupling_prepointamp,p[is].pre,sizeof(*p[is].pre)*PACKETBLOBS);
+    memcpy(g->coupling_postpointamp,p[is].post,sizeof(*p[is].post)*PACKETBLOBS);
+
+    if(hi->managed){
+      /* interpolate the kHz threshholds */
+      for(i=0;i<PACKETBLOBS;i++){
+        float kHz=p[is].kHz[i]*(1.-ds)+p[is+1].kHz[i]*ds;
+        g->coupling_pointlimit[0][i]=kHz*1000./vi->rate*ci->blocksizes[0];
+        g->coupling_pointlimit[1][i]=kHz*1000./vi->rate*ci->blocksizes[1];
+        g->coupling_pkHz[i]=kHz;
+
+        kHz=p[is].lowpasskHz[i]*(1.-ds)+p[is+1].lowpasskHz[i]*ds;
+        g->sliding_lowpass[0][i]=kHz*1000./vi->rate*ci->blocksizes[0];
+        g->sliding_lowpass[1][i]=kHz*1000./vi->rate*ci->blocksizes[1];
+
+      }
+    }else{
+      float kHz=p[is].kHz[PACKETBLOBS/2]*(1.-ds)+p[is+1].kHz[PACKETBLOBS/2]*ds;
+      for(i=0;i<PACKETBLOBS;i++){
+        g->coupling_pointlimit[0][i]=kHz*1000./vi->rate*ci->blocksizes[0];
+        g->coupling_pointlimit[1][i]=kHz*1000./vi->rate*ci->blocksizes[1];
+        g->coupling_pkHz[i]=kHz;
+      }
+
+      kHz=p[is].lowpasskHz[PACKETBLOBS/2]*(1.-ds)+p[is+1].lowpasskHz[PACKETBLOBS/2]*ds;
+      for(i=0;i<PACKETBLOBS;i++){
+        g->sliding_lowpass[0][i]=kHz*1000./vi->rate*ci->blocksizes[0];
+        g->sliding_lowpass[1][i]=kHz*1000./vi->rate*ci->blocksizes[1];
+      }
+    }
+  }else{
+    for(i=0;i<PACKETBLOBS;i++){
+      g->sliding_lowpass[0][i]=ci->blocksizes[0];
+      g->sliding_lowpass[1][i]=ci->blocksizes[1];
+    }
+  }
+  return;
+}
+
+static void vorbis_encode_psyset_setup(vorbis_info *vi,double s,
+                                       const int *nn_start,
+                                       const int *nn_partition,
+                                       const double *nn_thresh,
+                                       int block){
+  codec_setup_info *ci=vi->codec_setup;
+  vorbis_info_psy *p=ci->psy_param[block];
+  highlevel_encode_setup *hi=&ci->hi;
+  int is=s;
+
+  if(block>=ci->psys)
+    ci->psys=block+1;
+  if(!p){
+    p=_ogg_calloc(1,sizeof(*p));
+    ci->psy_param[block]=p;
+  }
+
+  memcpy(p,&_psy_info_template,sizeof(*p));
+  p->blockflag=block>>1;
+
+  if(hi->noise_normalize_p){
+    p->normal_p=1;
+    p->normal_start=nn_start[is];
+    p->normal_partition=nn_partition[is];
+    p->normal_thresh=nn_thresh[is];
+  }
+
+  return;
+}
+
+static void vorbis_encode_tonemask_setup(vorbis_info *vi,double s,int block,
+                                         const att3 *att,
+                                         const int  *max,
+                                         const vp_adjblock *in){
+  int i,is=s;
+  double ds=s-is;
+  codec_setup_info *ci=vi->codec_setup;
+  vorbis_info_psy *p=ci->psy_param[block];
+
+  /* 0 and 2 are only used by bitmanagement, but there's no harm to always
+     filling the values in here */
+  p->tone_masteratt[0]=att[is].att[0]*(1.-ds)+att[is+1].att[0]*ds;
+  p->tone_masteratt[1]=att[is].att[1]*(1.-ds)+att[is+1].att[1]*ds;
+  p->tone_masteratt[2]=att[is].att[2]*(1.-ds)+att[is+1].att[2]*ds;
+  p->tone_centerboost=att[is].boost*(1.-ds)+att[is+1].boost*ds;
+  p->tone_decay=att[is].decay*(1.-ds)+att[is+1].decay*ds;
+
+  p->max_curve_dB=max[is]*(1.-ds)+max[is+1]*ds;
+
+  for(i=0;i<P_BANDS;i++)
+    p->toneatt[i]=in[is].block[i]*(1.-ds)+in[is+1].block[i]*ds;
+  return;
+}
+
+
+static void vorbis_encode_compand_setup(vorbis_info *vi,double s,int block,
+                                        const compandblock *in,
+                                        const double *x){
+  int i,is=s;
+  double ds=s-is;
+  codec_setup_info *ci=vi->codec_setup;
+  vorbis_info_psy *p=ci->psy_param[block];
+
+  ds=x[is]*(1.-ds)+x[is+1]*ds;
+  is=(int)ds;
+  ds-=is;
+  if(ds==0 && is>0){
+    is--;
+    ds=1.;
+  }
+
+  /* interpolate the compander settings */
+  for(i=0;i<NOISE_COMPAND_LEVELS;i++)
+    p->noisecompand[i]=in[is].data[i]*(1.-ds)+in[is+1].data[i]*ds;
+  return;
+}
+
+static void vorbis_encode_peak_setup(vorbis_info *vi,double s,int block,
+                                    const int *suppress){
+  int is=s;
+  double ds=s-is;
+  codec_setup_info *ci=vi->codec_setup;
+  vorbis_info_psy *p=ci->psy_param[block];
+
+  p->tone_abs_limit=suppress[is]*(1.-ds)+suppress[is+1]*ds;
+
+  return;
+}
+
+static void vorbis_encode_noisebias_setup(vorbis_info *vi,double s,int block,
+                                         const int *suppress,
+                                         const noise3 *in,
+                                         const noiseguard *guard,
+                                         double userbias){
+  int i,is=s,j;
+  double ds=s-is;
+  codec_setup_info *ci=vi->codec_setup;
+  vorbis_info_psy *p=ci->psy_param[block];
+
+  p->noisemaxsupp=suppress[is]*(1.-ds)+suppress[is+1]*ds;
+  p->noisewindowlomin=guard[block].lo;
+  p->noisewindowhimin=guard[block].hi;
+  p->noisewindowfixed=guard[block].fixed;
+
+  for(j=0;j<P_NOISECURVES;j++)
+    for(i=0;i<P_BANDS;i++)
+      p->noiseoff[j][i]=in[is].data[j][i]*(1.-ds)+in[is+1].data[j][i]*ds;
+
+  /* impulse blocks may take a user specified bias to boost the
+     nominal/high noise encoding depth */
+  for(j=0;j<P_NOISECURVES;j++){
+    float min=p->noiseoff[j][0]+6; /* the lowest it can go */
+    for(i=0;i<P_BANDS;i++){
+      p->noiseoff[j][i]+=userbias;
+      if(p->noiseoff[j][i]<min)p->noiseoff[j][i]=min;
+    }
+  }
+
+  return;
+}
+
+static void vorbis_encode_ath_setup(vorbis_info *vi,int block){
+  codec_setup_info *ci=vi->codec_setup;
+  vorbis_info_psy *p=ci->psy_param[block];
+
+  p->ath_adjatt=ci->hi.ath_floating_dB;
+  p->ath_maxatt=ci->hi.ath_absolute_dB;
+  return;
+}
+
+
+static int book_dup_or_new(codec_setup_info *ci,const static_codebook *book){
+  int i;
+  for(i=0;i<ci->books;i++)
+    if(ci->book_param[i]==book)return(i);
+
+  return(ci->books++);
+}
+
+static void vorbis_encode_blocksize_setup(vorbis_info *vi,double s,
+                                         const int *shortb,const int *longb){
+
+  codec_setup_info *ci=vi->codec_setup;
+  int is=s;
+
+  int blockshort=shortb[is];
+  int blocklong=longb[is];
+  ci->blocksizes[0]=blockshort;
+  ci->blocksizes[1]=blocklong;
+
+}
+
+static void vorbis_encode_residue_setup(vorbis_info *vi,
+                                        int number, int block,
+                                        const vorbis_residue_template *res){
+
+  codec_setup_info *ci=vi->codec_setup;
+  int i;
+
+  vorbis_info_residue0 *r=ci->residue_param[number]=
+    _ogg_malloc(sizeof(*r));
+
+  memcpy(r,res->res,sizeof(*r));
+  if(ci->residues<=number)ci->residues=number+1;
+
+  r->grouping=res->grouping;
+  ci->residue_type[number]=res->res_type;
+
+  /* fill in all the books */
+  {
+    int booklist=0,k;
+
+    if(ci->hi.managed){
+      for(i=0;i<r->partitions;i++)
+        for(k=0;k<4;k++)
+          if(res->books_base_managed->books[i][k])
+            r->secondstages[i]|=(1<<k);
+
+      r->groupbook=book_dup_or_new(ci,res->book_aux_managed);
+      ci->book_param[r->groupbook]=(static_codebook *)res->book_aux_managed;
+
+      for(i=0;i<r->partitions;i++){
+        for(k=0;k<4;k++){
+          if(res->books_base_managed->books[i][k]){
+            int bookid=book_dup_or_new(ci,res->books_base_managed->books[i][k]);
+            r->booklist[booklist++]=bookid;
+            ci->book_param[bookid]=(static_codebook *)res->books_base_managed->books[i][k];
+          }
+        }
+      }
+
+    }else{
+
+      for(i=0;i<r->partitions;i++)
+        for(k=0;k<4;k++)
+          if(res->books_base->books[i][k])
+            r->secondstages[i]|=(1<<k);
+
+      r->groupbook=book_dup_or_new(ci,res->book_aux);
+      ci->book_param[r->groupbook]=(static_codebook *)res->book_aux;
+
+      for(i=0;i<r->partitions;i++){
+        for(k=0;k<4;k++){
+          if(res->books_base->books[i][k]){
+            int bookid=book_dup_or_new(ci,res->books_base->books[i][k]);
+            r->booklist[booklist++]=bookid;
+            ci->book_param[bookid]=(static_codebook *)res->books_base->books[i][k];
+          }
+        }
+      }
+    }
+  }
+
+  /* lowpass setup/pointlimit */
+  {
+    double freq=ci->hi.lowpass_kHz*1000.;
+    vorbis_info_floor1 *f=ci->floor_param[block]; /* by convention */
+    double nyq=vi->rate/2.;
+    long blocksize=ci->blocksizes[block]>>1;
+
+    /* lowpass needs to be set in the floor and the residue. */
+    if(freq>nyq)freq=nyq;
+    /* in the floor, the granularity can be very fine; it doesn't alter
+       the encoding structure, only the samples used to fit the floor
+       approximation */
+    f->n=freq/nyq*blocksize;
+
+    /* this res may by limited by the maximum pointlimit of the mode,
+       not the lowpass. the floor is always lowpass limited. */
+    switch(res->limit_type){
+    case 1: /* point stereo limited */
+      if(ci->hi.managed)
+        freq=ci->psy_g_param.coupling_pkHz[PACKETBLOBS-1]*1000.;
+      else
+        freq=ci->psy_g_param.coupling_pkHz[PACKETBLOBS/2]*1000.;
+      if(freq>nyq)freq=nyq;
+      break;
+    case 2: /* LFE channel; lowpass at ~ 250Hz */
+      freq=250;
+      break;
+    default:
+      /* already set */
+      break;
+    }
+
+    /* in the residue, we're constrained, physically, by partition
+       boundaries.  We still lowpass 'wherever', but we have to round up
+       here to next boundary, or the vorbis spec will round it *down* to
+       previous boundary in encode/decode */
+    if(ci->residue_type[number]==2){
+      /* residue 2 bundles together multiple channels; used by stereo
+         and surround.  Count the channels in use */
+      /* Multiple maps/submaps can point to the same residue.  In the case
+         of residue 2, they all better have the same number of
+         channels/samples. */
+      int j,k,ch=0;
+      for(i=0;i<ci->maps&&ch==0;i++){
+        vorbis_info_mapping0 *mi=(vorbis_info_mapping0 *)ci->map_param[i];
+        for(j=0;j<mi->submaps && ch==0;j++)
+          if(mi->residuesubmap[j]==number) /* we found a submap referencing theis residue backend */
+            for(k=0;k<vi->channels;k++)
+              if(mi->chmuxlist[k]==j) /* this channel belongs to the submap */
+                ch++;
+      }
+
+      r->end=(int)((freq/nyq*blocksize*ch)/r->grouping+.9)* /* round up only if we're well past */
+        r->grouping;
+      /* the blocksize and grouping may disagree at the end */
+      if(r->end>blocksize*ch)r->end=blocksize*ch/r->grouping*r->grouping;
+
+    }else{
+
+      r->end=(int)((freq/nyq*blocksize)/r->grouping+.9)* /* round up only if we're well past */
+        r->grouping;
+      /* the blocksize and grouping may disagree at the end */
+      if(r->end>blocksize)r->end=blocksize/r->grouping*r->grouping;
+
+    }
+
+    if(r->end==0)r->end=r->grouping; /* LFE channel */
+
+  }
+}
+
+/* we assume two maps in this encoder */
+static void vorbis_encode_map_n_res_setup(vorbis_info *vi,double s,
+                                          const vorbis_mapping_template *maps){
+
+  codec_setup_info *ci=vi->codec_setup;
+  int i,j,is=s,modes=2;
+  const vorbis_info_mapping0 *map=maps[is].map;
+  const vorbis_info_mode *mode=_mode_template;
+  const vorbis_residue_template *res=maps[is].res;
+
+  if(ci->blocksizes[0]==ci->blocksizes[1])modes=1;
+
+  for(i=0;i<modes;i++){
+
+    ci->map_param[i]=_ogg_calloc(1,sizeof(*map));
+    ci->mode_param[i]=_ogg_calloc(1,sizeof(*mode));
+
+    memcpy(ci->mode_param[i],mode+i,sizeof(*_mode_template));
+    if(i>=ci->modes)ci->modes=i+1;
+
+    ci->map_type[i]=0;
+    memcpy(ci->map_param[i],map+i,sizeof(*map));
+    if(i>=ci->maps)ci->maps=i+1;
+
+    for(j=0;j<map[i].submaps;j++)
+      vorbis_encode_residue_setup(vi,map[i].residuesubmap[j],i
+                                  ,res+map[i].residuesubmap[j]);
+  }
+}
+
+static double setting_to_approx_bitrate(vorbis_info *vi){
+  codec_setup_info *ci=vi->codec_setup;
+  highlevel_encode_setup *hi=&ci->hi;
+  ve_setup_data_template *setup=(ve_setup_data_template *)hi->setup;
+  int is=hi->base_setting;
+  double ds=hi->base_setting-is;
+  int ch=vi->channels;
+  const double *r=setup->rate_mapping;
+
+  if(r==NULL)
+    return(-1);
+
+  return((r[is]*(1.-ds)+r[is+1]*ds)*ch);
+}
+
+static const void *get_setup_template(long ch,long srate,
+                                      double req,int q_or_bitrate,
+                                      double *base_setting){
+  int i=0,j;
+  if(q_or_bitrate)req/=ch;
+
+  while(setup_list[i]){
+    if(setup_list[i]->coupling_restriction==-1 ||
+       setup_list[i]->coupling_restriction==ch){
+      if(srate>=setup_list[i]->samplerate_min_restriction &&
+         srate<=setup_list[i]->samplerate_max_restriction){
+        int mappings=setup_list[i]->mappings;
+        const double *map=(q_or_bitrate?
+                     setup_list[i]->rate_mapping:
+                     setup_list[i]->quality_mapping);
+
+        /* the template matches.  Does the requested quality mode
+           fall within this template's modes? */
+        if(req<map[0]){++i;continue;}
+        if(req>map[setup_list[i]->mappings]){++i;continue;}
+        for(j=0;j<mappings;j++)
+          if(req>=map[j] && req<map[j+1])break;
+        /* an all-points match */
+        if(j==mappings)
+          *base_setting=j-.001;
+        else{
+          float low=map[j];
+          float high=map[j+1];
+          float del=(req-low)/(high-low);
+          *base_setting=j+del;
+        }
+
+        return(setup_list[i]);
+      }
+    }
+    i++;
+  }
+
+  return NULL;
+}
+
+/* encoders will need to use vorbis_info_init beforehand and call
+   vorbis_info clear when all done */
+
+/* two interfaces; this, more detailed one, and later a convenience
+   layer on top */
+
+/* the final setup call */
+int vorbis_encode_setup_init(vorbis_info *vi){
+  int i,i0=0,singleblock=0;
+  codec_setup_info *ci=vi->codec_setup;
+  ve_setup_data_template *setup=NULL;
+  highlevel_encode_setup *hi=&ci->hi;
+
+  if(ci==NULL)return(OV_EINVAL);
+  if(!hi->impulse_block_p)i0=1;
+
+  /* too low/high an ATH floater is nonsensical, but doesn't break anything */
+  if(hi->ath_floating_dB>-80)hi->ath_floating_dB=-80;
+  if(hi->ath_floating_dB<-200)hi->ath_floating_dB=-200;
+
+  /* again, bound this to avoid the app shooting itself int he foot
+     too badly */
+  if(hi->amplitude_track_dBpersec>0.)hi->amplitude_track_dBpersec=0.;
+  if(hi->amplitude_track_dBpersec<-99999.)hi->amplitude_track_dBpersec=-99999.;
+
+  /* get the appropriate setup template; matches the fetch in previous
+     stages */
+  setup=(ve_setup_data_template *)hi->setup;
+  if(setup==NULL)return(OV_EINVAL);
+
+  hi->set_in_stone=1;
+  /* choose block sizes from configured sizes as well as paying
+     attention to long_block_p and short_block_p.  If the configured
+     short and long blocks are the same length, we set long_block_p
+     and unset short_block_p */
+  vorbis_encode_blocksize_setup(vi,hi->base_setting,
+                                setup->blocksize_short,
+                                setup->blocksize_long);
+  if(ci->blocksizes[0]==ci->blocksizes[1])singleblock=1;
+
+  /* floor setup; choose proper floor params.  Allocated on the floor
+     stack in order; if we alloc only a single long floor, it's 0 */
+  for(i=0;i<setup->floor_mappings;i++)
+    vorbis_encode_floor_setup(vi,hi->base_setting,
+                              setup->floor_books,
+                              setup->floor_params,
+                              setup->floor_mapping_list[i]);
+
+  /* setup of [mostly] short block detection and stereo*/
+  vorbis_encode_global_psych_setup(vi,hi->trigger_setting,
+                                   setup->global_params,
+                                   setup->global_mapping);
+  vorbis_encode_global_stereo(vi,hi,setup->stereo_modes);
+
+  /* basic psych setup and noise normalization */
+  vorbis_encode_psyset_setup(vi,hi->base_setting,
+                             setup->psy_noise_normal_start[0],
+                             setup->psy_noise_normal_partition[0],
+                             setup->psy_noise_normal_thresh,
+                             0);
+  vorbis_encode_psyset_setup(vi,hi->base_setting,
+                             setup->psy_noise_normal_start[0],
+                             setup->psy_noise_normal_partition[0],
+                             setup->psy_noise_normal_thresh,
+                             1);
+  if(!singleblock){
+    vorbis_encode_psyset_setup(vi,hi->base_setting,
+                               setup->psy_noise_normal_start[1],
+                               setup->psy_noise_normal_partition[1],
+                                    setup->psy_noise_normal_thresh,
+                               2);
+    vorbis_encode_psyset_setup(vi,hi->base_setting,
+                               setup->psy_noise_normal_start[1],
+                               setup->psy_noise_normal_partition[1],
+                               setup->psy_noise_normal_thresh,
+                               3);
+  }
+
+  /* tone masking setup */
+  vorbis_encode_tonemask_setup(vi,hi->block[i0].tone_mask_setting,0,
+                               setup->psy_tone_masteratt,
+                               setup->psy_tone_0dB,
+                               setup->psy_tone_adj_impulse);
+  vorbis_encode_tonemask_setup(vi,hi->block[1].tone_mask_setting,1,
+                               setup->psy_tone_masteratt,
+                               setup->psy_tone_0dB,
+                               setup->psy_tone_adj_other);
+  if(!singleblock){
+    vorbis_encode_tonemask_setup(vi,hi->block[2].tone_mask_setting,2,
+                                 setup->psy_tone_masteratt,
+                                 setup->psy_tone_0dB,
+                                 setup->psy_tone_adj_other);
+    vorbis_encode_tonemask_setup(vi,hi->block[3].tone_mask_setting,3,
+                                 setup->psy_tone_masteratt,
+                                 setup->psy_tone_0dB,
+                                 setup->psy_tone_adj_long);
+  }
+
+  /* noise companding setup */
+  vorbis_encode_compand_setup(vi,hi->block[i0].noise_compand_setting,0,
+                              setup->psy_noise_compand,
+                              setup->psy_noise_compand_short_mapping);
+  vorbis_encode_compand_setup(vi,hi->block[1].noise_compand_setting,1,
+                              setup->psy_noise_compand,
+                              setup->psy_noise_compand_short_mapping);
+  if(!singleblock){
+    vorbis_encode_compand_setup(vi,hi->block[2].noise_compand_setting,2,
+                                setup->psy_noise_compand,
+                                setup->psy_noise_compand_long_mapping);
+    vorbis_encode_compand_setup(vi,hi->block[3].noise_compand_setting,3,
+                                setup->psy_noise_compand,
+                                setup->psy_noise_compand_long_mapping);
+  }
+
+  /* peak guarding setup  */
+  vorbis_encode_peak_setup(vi,hi->block[i0].tone_peaklimit_setting,0,
+                           setup->psy_tone_dBsuppress);
+  vorbis_encode_peak_setup(vi,hi->block[1].tone_peaklimit_setting,1,
+                           setup->psy_tone_dBsuppress);
+  if(!singleblock){
+    vorbis_encode_peak_setup(vi,hi->block[2].tone_peaklimit_setting,2,
+                             setup->psy_tone_dBsuppress);
+    vorbis_encode_peak_setup(vi,hi->block[3].tone_peaklimit_setting,3,
+                             setup->psy_tone_dBsuppress);
+  }
+
+  /* noise bias setup */
+  vorbis_encode_noisebias_setup(vi,hi->block[i0].noise_bias_setting,0,
+                                setup->psy_noise_dBsuppress,
+                                setup->psy_noise_bias_impulse,
+                                setup->psy_noiseguards,
+                                (i0==0?hi->impulse_noisetune:0.));
+  vorbis_encode_noisebias_setup(vi,hi->block[1].noise_bias_setting,1,
+                                setup->psy_noise_dBsuppress,
+                                setup->psy_noise_bias_padding,
+                                setup->psy_noiseguards,0.);
+  if(!singleblock){
+    vorbis_encode_noisebias_setup(vi,hi->block[2].noise_bias_setting,2,
+                                  setup->psy_noise_dBsuppress,
+                                  setup->psy_noise_bias_trans,
+                                  setup->psy_noiseguards,0.);
+    vorbis_encode_noisebias_setup(vi,hi->block[3].noise_bias_setting,3,
+                                  setup->psy_noise_dBsuppress,
+                                  setup->psy_noise_bias_long,
+                                  setup->psy_noiseguards,0.);
+  }
+
+  vorbis_encode_ath_setup(vi,0);
+  vorbis_encode_ath_setup(vi,1);
+  if(!singleblock){
+    vorbis_encode_ath_setup(vi,2);
+    vorbis_encode_ath_setup(vi,3);
+  }
+
+  vorbis_encode_map_n_res_setup(vi,hi->base_setting,setup->maps);
+
+  /* set bitrate readonlies and management */
+  if(hi->bitrate_av>0)
+    vi->bitrate_nominal=hi->bitrate_av;
+  else{
+    vi->bitrate_nominal=setting_to_approx_bitrate(vi);
+  }
+
+  vi->bitrate_lower=hi->bitrate_min;
+  vi->bitrate_upper=hi->bitrate_max;
+  if(hi->bitrate_av)
+    vi->bitrate_window=(double)hi->bitrate_reservoir/hi->bitrate_av;
+  else
+    vi->bitrate_window=0.;
+
+  if(hi->managed){
+    ci->bi.avg_rate=hi->bitrate_av;
+    ci->bi.min_rate=hi->bitrate_min;
+    ci->bi.max_rate=hi->bitrate_max;
+
+    ci->bi.reservoir_bits=hi->bitrate_reservoir;
+    ci->bi.reservoir_bias=
+      hi->bitrate_reservoir_bias;
+
+    ci->bi.slew_damp=hi->bitrate_av_damp;
+
+  }
+
+  return(0);
+
+}
+
+static void vorbis_encode_setup_setting(vorbis_info *vi,
+                                       long  channels,
+                                       long  rate){
+  int i,is;
+  codec_setup_info *ci=vi->codec_setup;
+  highlevel_encode_setup *hi=&ci->hi;
+  const ve_setup_data_template *setup=hi->setup;
+  double ds;
+
+  vi->version=0;
+  vi->channels=channels;
+  vi->rate=rate;
+
+  hi->impulse_block_p=1;
+  hi->noise_normalize_p=1;
+
+  is=hi->base_setting;
+  ds=hi->base_setting-is;
+
+  hi->stereo_point_setting=hi->base_setting;
+
+  if(!hi->lowpass_altered)
+    hi->lowpass_kHz=
+      setup->psy_lowpass[is]*(1.-ds)+setup->psy_lowpass[is+1]*ds;
+
+  hi->ath_floating_dB=setup->psy_ath_float[is]*(1.-ds)+
+    setup->psy_ath_float[is+1]*ds;
+  hi->ath_absolute_dB=setup->psy_ath_abs[is]*(1.-ds)+
+    setup->psy_ath_abs[is+1]*ds;
+
+  hi->amplitude_track_dBpersec=-6.;
+  hi->trigger_setting=hi->base_setting;
+
+  for(i=0;i<4;i++){
+    hi->block[i].tone_mask_setting=hi->base_setting;
+    hi->block[i].tone_peaklimit_setting=hi->base_setting;
+    hi->block[i].noise_bias_setting=hi->base_setting;
+    hi->block[i].noise_compand_setting=hi->base_setting;
+  }
+}
+
+int vorbis_encode_setup_vbr(vorbis_info *vi,
+                            long  channels,
+                            long  rate,
+                            float quality){
+  codec_setup_info *ci=vi->codec_setup;
+  highlevel_encode_setup *hi=&ci->hi;
+
+  quality+=.0000001;
+  if(quality>=1.)quality=.9999;
+
+  hi->req=quality;
+  hi->setup=get_setup_template(channels,rate,quality,0,&hi->base_setting);
+  if(!hi->setup)return OV_EIMPL;
+
+  vorbis_encode_setup_setting(vi,channels,rate);
+  hi->managed=0;
+  hi->coupling_p=1;
+
+  return 0;
+}
+
+int vorbis_encode_init_vbr(vorbis_info *vi,
+                           long channels,
+                           long rate,
+
+                           float base_quality /* 0. to 1. */
+                           ){
+  int ret=0;
+
+  ret=vorbis_encode_setup_vbr(vi,channels,rate,base_quality);
+
+  if(ret){
+    vorbis_info_clear(vi);
+    return ret;
+  }
+  ret=vorbis_encode_setup_init(vi);
+  if(ret)
+    vorbis_info_clear(vi);
+  return(ret);
+}
+
+int vorbis_encode_setup_managed(vorbis_info *vi,
+                                long channels,
+                                long rate,
+
+                                long max_bitrate,
+                                long nominal_bitrate,
+                                long min_bitrate){
+
+  codec_setup_info *ci=vi->codec_setup;
+  highlevel_encode_setup *hi=&ci->hi;
+  double tnominal=nominal_bitrate;
+
+  if(nominal_bitrate<=0.){
+    if(max_bitrate>0.){
+      if(min_bitrate>0.)
+        nominal_bitrate=(max_bitrate+min_bitrate)*.5;
+      else
+        nominal_bitrate=max_bitrate*.875;
+    }else{
+      if(min_bitrate>0.){
+        nominal_bitrate=min_bitrate;
+      }else{
+        return(OV_EINVAL);
+      }
+    }
+  }
+
+  hi->req=nominal_bitrate;
+  hi->setup=get_setup_template(channels,rate,nominal_bitrate,1,&hi->base_setting);
+  if(!hi->setup)return OV_EIMPL;
+
+  vorbis_encode_setup_setting(vi,channels,rate);
+
+  /* initialize management with sane defaults */
+  hi->coupling_p=1;
+  hi->managed=1;
+  hi->bitrate_min=min_bitrate;
+  hi->bitrate_max=max_bitrate;
+  hi->bitrate_av=tnominal;
+  hi->bitrate_av_damp=1.5f; /* full range in no less than 1.5 second */
+  hi->bitrate_reservoir=nominal_bitrate*2;
+  hi->bitrate_reservoir_bias=.1; /* bias toward hoarding bits */
+
+  return(0);
+
+}
+
+int vorbis_encode_init(vorbis_info *vi,
+                       long channels,
+                       long rate,
+
+                       long max_bitrate,
+                       long nominal_bitrate,
+                       long min_bitrate){
+
+  int ret=vorbis_encode_setup_managed(vi,channels,rate,
+                                      max_bitrate,
+                                      nominal_bitrate,
+                                      min_bitrate);
+  if(ret){
+    vorbis_info_clear(vi);
+    return(ret);
+  }
+
+  ret=vorbis_encode_setup_init(vi);
+  if(ret)
+    vorbis_info_clear(vi);
+  return(ret);
+}
+
+int vorbis_encode_ctl(vorbis_info *vi,int number,void *arg){
+  if(vi){
+    codec_setup_info *ci=vi->codec_setup;
+    highlevel_encode_setup *hi=&ci->hi;
+    int setp=(number&0xf); /* a read request has a low nibble of 0 */
+
+    if(setp && hi->set_in_stone)return(OV_EINVAL);
+
+    switch(number){
+
+    /* now deprecated *****************/
+    case OV_ECTL_RATEMANAGE_GET:
+      {
+
+        struct ovectl_ratemanage_arg *ai=
+          (struct ovectl_ratemanage_arg *)arg;
+
+        ai->management_active=hi->managed;
+        ai->bitrate_hard_window=ai->bitrate_av_window=
+          (double)hi->bitrate_reservoir/vi->rate;
+        ai->bitrate_av_window_center=1.;
+        ai->bitrate_hard_min=hi->bitrate_min;
+        ai->bitrate_hard_max=hi->bitrate_max;
+        ai->bitrate_av_lo=hi->bitrate_av;
+        ai->bitrate_av_hi=hi->bitrate_av;
+
+      }
+      return(0);
+
+    /* now deprecated *****************/
+    case OV_ECTL_RATEMANAGE_SET:
+      {
+        struct ovectl_ratemanage_arg *ai=
+          (struct ovectl_ratemanage_arg *)arg;
+        if(ai==NULL){
+          hi->managed=0;
+        }else{
+          hi->managed=ai->management_active;
+          vorbis_encode_ctl(vi,OV_ECTL_RATEMANAGE_AVG,arg);
+          vorbis_encode_ctl(vi,OV_ECTL_RATEMANAGE_HARD,arg);
+        }
+      }
+      return 0;
+
+    /* now deprecated *****************/
+    case OV_ECTL_RATEMANAGE_AVG:
+      {
+        struct ovectl_ratemanage_arg *ai=
+          (struct ovectl_ratemanage_arg *)arg;
+        if(ai==NULL){
+          hi->bitrate_av=0;
+        }else{
+          hi->bitrate_av=(ai->bitrate_av_lo+ai->bitrate_av_hi)*.5;
+        }
+      }
+      return(0);
+    /* now deprecated *****************/
+    case OV_ECTL_RATEMANAGE_HARD:
+      {
+        struct ovectl_ratemanage_arg *ai=
+          (struct ovectl_ratemanage_arg *)arg;
+        if(ai==NULL){
+          hi->bitrate_min=0;
+          hi->bitrate_max=0;
+        }else{
+          hi->bitrate_min=ai->bitrate_hard_min;
+          hi->bitrate_max=ai->bitrate_hard_max;
+          hi->bitrate_reservoir=ai->bitrate_hard_window*
+            (hi->bitrate_max+hi->bitrate_min)*.5;
+        }
+        if(hi->bitrate_reservoir<128.)
+          hi->bitrate_reservoir=128.;
+      }
+      return(0);
+
+      /* replacement ratemanage interface */
+    case OV_ECTL_RATEMANAGE2_GET:
+      {
+        struct ovectl_ratemanage2_arg *ai=
+          (struct ovectl_ratemanage2_arg *)arg;
+        if(ai==NULL)return OV_EINVAL;
+
+        ai->management_active=hi->managed;
+        ai->bitrate_limit_min_kbps=hi->bitrate_min/1000;
+        ai->bitrate_limit_max_kbps=hi->bitrate_max/1000;
+        ai->bitrate_average_kbps=hi->bitrate_av/1000;
+        ai->bitrate_average_damping=hi->bitrate_av_damp;
+        ai->bitrate_limit_reservoir_bits=hi->bitrate_reservoir;
+        ai->bitrate_limit_reservoir_bias=hi->bitrate_reservoir_bias;
+      }
+      return (0);
+    case OV_ECTL_RATEMANAGE2_SET:
+      {
+        struct ovectl_ratemanage2_arg *ai=
+          (struct ovectl_ratemanage2_arg *)arg;
+        if(ai==NULL){
+          hi->managed=0;
+        }else{
+          /* sanity check; only catch invariant violations */
+          if(ai->bitrate_limit_min_kbps>0 &&
+             ai->bitrate_average_kbps>0 &&
+             ai->bitrate_limit_min_kbps>ai->bitrate_average_kbps)
+            return OV_EINVAL;
+
+          if(ai->bitrate_limit_max_kbps>0 &&
+             ai->bitrate_average_kbps>0 &&
+             ai->bitrate_limit_max_kbps<ai->bitrate_average_kbps)
+            return OV_EINVAL;
+
+          if(ai->bitrate_limit_min_kbps>0 &&
+             ai->bitrate_limit_max_kbps>0 &&
+             ai->bitrate_limit_min_kbps>ai->bitrate_limit_max_kbps)
+            return OV_EINVAL;
+
+          if(ai->bitrate_average_damping <= 0.)
+            return OV_EINVAL;
+
+          if(ai->bitrate_limit_reservoir_bits < 0)
+            return OV_EINVAL;
+
+          if(ai->bitrate_limit_reservoir_bias < 0.)
+            return OV_EINVAL;
+
+          if(ai->bitrate_limit_reservoir_bias > 1.)
+            return OV_EINVAL;
+
+          hi->managed=ai->management_active;
+          hi->bitrate_min=ai->bitrate_limit_min_kbps * 1000;
+          hi->bitrate_max=ai->bitrate_limit_max_kbps * 1000;
+          hi->bitrate_av=ai->bitrate_average_kbps * 1000;
+          hi->bitrate_av_damp=ai->bitrate_average_damping;
+          hi->bitrate_reservoir=ai->bitrate_limit_reservoir_bits;
+          hi->bitrate_reservoir_bias=ai->bitrate_limit_reservoir_bias;
+        }
+      }
+      return 0;
+
+    case OV_ECTL_LOWPASS_GET:
+      {
+        double *farg=(double *)arg;
+        *farg=hi->lowpass_kHz;
+      }
+      return(0);
+    case OV_ECTL_LOWPASS_SET:
+      {
+        double *farg=(double *)arg;
+        hi->lowpass_kHz=*farg;
+
+        if(hi->lowpass_kHz<2.)hi->lowpass_kHz=2.;
+        if(hi->lowpass_kHz>99.)hi->lowpass_kHz=99.;
+        hi->lowpass_altered=1;
+      }
+      return(0);
+    case OV_ECTL_IBLOCK_GET:
+      {
+        double *farg=(double *)arg;
+        *farg=hi->impulse_noisetune;
+      }
+      return(0);
+    case OV_ECTL_IBLOCK_SET:
+      {
+        double *farg=(double *)arg;
+        hi->impulse_noisetune=*farg;
+
+        if(hi->impulse_noisetune>0.)hi->impulse_noisetune=0.;
+        if(hi->impulse_noisetune<-15.)hi->impulse_noisetune=-15.;
+      }
+      return(0);
+    case OV_ECTL_COUPLING_GET:
+      {
+        int *iarg=(int *)arg;
+        *iarg=hi->coupling_p;
+      }
+      return(0);
+    case OV_ECTL_COUPLING_SET:
+      {
+        const void *new_template;
+        double new_base=0.;
+        int *iarg=(int *)arg;
+        hi->coupling_p=((*iarg)!=0);
+
+        /* Fetching a new template can alter the base_setting, which
+           many other parameters are based on.  Right now, the only
+           parameter drawn from the base_setting that can be altered
+           by an encctl is the lowpass, so that is explictly flagged
+           to not be overwritten when we fetch a new template and
+           recompute the dependant settings */
+        new_template = get_setup_template(hi->coupling_p?vi->channels:-1,
+                                          vi->rate,
+                                          hi->req,
+                                          hi->managed,
+                                          &new_base);
+        if(!hi->setup)return OV_EIMPL;
+        hi->setup=new_template;
+        hi->base_setting=new_base;
+        vorbis_encode_setup_setting(vi,vi->channels,vi->rate);
+      }
+      return(0);
+    }
+    return(OV_EIMPL);
+  }
+  return(OV_EINVAL);
+}