ref: a9411472cd82b466dc786d71a10120e919f4d8c6
parent: eedb42282acee422d311919c2979a2353cd15a22
author: Gregory Maxwell <[email protected]>
date: Wed Oct 27 23:52:21 EDT 2010
Switch example tools to use VBR and 960 sample frames by default on the basis that if the user doesn't have any particular requirements that they probably want this. Minor change in the VBR behavior to hot-start with some internal state parameters which were observed to be quite consistent across bitrates, framesizes, and content. This also prevents it from completely burning the reserve capacity on the first frame if its a short. Also switch some maximum frame sizes to match the OPUS draft maximums.
--- a/libcelt/celt.c
+++ b/libcelt/celt.c
@@ -67,7 +67,7 @@
int complexity;
int start, end;
- celt_int32 vbr_rate_norm; /* Target number of 16th bits per frame */
+ celt_int32 vbr_rate_norm; /* Target number of 8th bits per frame */
/* Everything beyond this point gets cleared on a reset */
#define ENCODER_RESET_START frame_max
@@ -132,6 +132,7 @@
st->end = st->mode->effEBands;
st->vbr_rate_norm = 0;
+ st->vbr_offset = -140<<BITRES;
st->force_intra = 0;
st->delayedIntra = 1;
st->tonal_average = 256;
@@ -794,7 +795,7 @@
{celt_word16 alpha;
celt_int32 delta;
- /* The target rate in 16th bits per frame */
+ /* The target rate in 8th bits per frame */
celt_int32 vbr_rate;
celt_int32 target;
celt_int32 vbr_bound, max_allowed;
@@ -819,18 +820,16 @@
else if (M > 1)
target-=(target+14)/28;
- /* The average energy is removed from the target and the actual
- energy added*/
- target=target+st->vbr_offset-(50<<BITRES)+ec_enc_tell(enc, BITRES);
+ /* The current offset is removed from the target and the space used
+ so far is added*/
+ target=target+st->vbr_offset+ec_enc_tell(enc, BITRES);
/* In VBR mode the frame size must not be reduced so much that it would result in the coarse energy busting its budget */
target=IMIN(nbAvailableBytes<<(BITRES+3),target);
- /* Make the adaptation coef (alpha) higher at the beginning */
- if (st->vbr_count < 990)
+ if (st->vbr_count < 970)
{st->vbr_count++;
- alpha = celt_rcp(SHL32(EXTEND32(st->vbr_count+10),16));
- /*printf ("%d %d\n", st->vbr_count+10, alpha);*/+ alpha = celt_rcp(SHL32(EXTEND32(st->vbr_count+20),16));
} else
alpha = QCONST16(.001f,15);
--- a/libcelt/testcelt.c
+++ b/libcelt/testcelt.c
@@ -41,7 +41,7 @@
#include <math.h>
#include <string.h>
-#define MAX_PACKET 1024
+#define MAX_PACKET 1275
int main(int argc, char *argv[])
{--- a/tools/celtenc.c
+++ b/tools/celtenc.c
@@ -1,6 +1,6 @@
-/* Copyright (c) 2002-2007 Jean-Marc Valin
- Copyright (c) 2007-2009 Xiph.Org Foundation
- Copyright (c) 2008-2009 Gregory Maxwell
+/* Copyright (c) 2002-2010 Jean-Marc Valin
+ Copyright (c) 2007-2010 Xiph.Org Foundation
+ Copyright (c) 2008-2010 Gregory Maxwell
File: celtenc.c
Redistribution and use in source and binary forms, with or without
@@ -81,7 +81,7 @@
}
#define MAX_FRAME_SIZE 2048
-#define MAX_FRAME_BYTES 1024
+#define MAX_FRAME_BYTES 1275
#define IMIN(a,b) ((a) < (b) ? (a) : (b)) /**< Minimum int value. */
#define IMAX(a,b) ((a) > (b) ? (a) : (b)) /**< Maximum int value. */
@@ -215,9 +215,9 @@
printf ("\n"); printf ("Options:\n"); printf (" --bitrate n Encoding bit-rate in kbit/sec\n"); - printf (" --vbr Use variable bitrate encoding\n"); + printf (" --cbr Use constant bitrate encoding\n"); printf (" --comp n Encoding complexity (0-10)\n");- printf (" --framesize n Frame size (Default: 256)\n");+ printf (" --framesize n Frame size (Default: 960)\n"); printf (" --noltp Do not use long-term prediction\n"); printf (" --independent Encode frames independently (implies noltp)\n"); printf (" --skeleton Outputs ogg skeleton metadata (may cause incompatibilities)\n");@@ -236,7 +236,7 @@
printf (" --be Raw input is big-endian\n"); printf (" --8bit Raw input is 8-bit unsigned\n"); printf (" --16bit Raw input is 16-bit signed\n"); - printf ("Default raw PCM input is 16-bit, little-endian, mono\n"); + printf ("Default raw PCM input is 48kHz, 16-bit, little-endian, stereo\n");}
@@ -248,13 +248,13 @@
char *inFile, *outFile;
FILE *fin, *fout;
short input[MAX_FRAME_SIZE];
- celt_int32 frame_size = 256;
+ celt_int32 frame_size = 960;
int quiet=0;
int nbBytes;
CELTMode *mode;
void *st;
unsigned char bits[MAX_FRAME_BYTES];
- int with_vbr = 0;
+ int with_cbr = 0;
int with_skeleton = 0;
int total_bytes = 0;
int peak_bytes = 0;
@@ -261,7 +261,7 @@
struct option long_options[] =
{ {"bitrate", required_argument, NULL, 0},- {"vbr",no_argument,NULL, 0},+ {"cbr",no_argument,NULL, 0}, {"comp", required_argument, NULL, 0}, {"noltp", no_argument, NULL, 0}, {"independent", no_argument, NULL, 0},@@ -284,7 +284,7 @@
{0, 0, 0, 0}};
int print_bitrate=0;
- celt_int32 rate=44100;
+ celt_int32 rate=48000;
celt_int32 size;
int chan=1;
int fmt=16;
@@ -305,7 +305,7 @@
char first_bytes[12];
int wave_input=0;
celt_int32 lookahead = 0;
- int bytes_per_packet=48;
+ int bytes_per_packet=-1;
int complexity=-127;
int prediction=2;
int bitstream;
@@ -325,9 +325,9 @@
if (strcmp(long_options[option_index].name,"bitrate")==0)
{bitrate = atof (optarg);
- } else if (strcmp(long_options[option_index].name,"vbr")==0)
+ } else if (strcmp(long_options[option_index].name,"cbr")==0)
{- with_vbr=1;
+ with_cbr=1;
} else if (strcmp(long_options[option_index].name,"skeleton")==0)
{with_skeleton=1;
@@ -483,12 +483,12 @@
fprintf (stderr, "Warning: Requested bitrate (%0.3fkbit/sec) is too high. Setting CELT to %d bytes/frame.\n",bitrate,MAX_FRAME_BYTES);
}
- if (with_vbr)
+ if (with_cbr)
{+ bitrate = ((rate/(float)frame_size)*8*bytes_per_packet)/1000.0;
+ } else {/*In VBR mode the bytes_per_packet argument becomes a hard maximum. 3x the average rate is just a random choice.*/
bytes_per_packet=IMIN(bytes_per_packet*3,MAX_FRAME_BYTES);
- } else { - bitrate = ((rate/(float)frame_size)*8*bytes_per_packet)/1000.0;
}
mode = celt_mode_create(rate, frame_size, NULL);
@@ -510,18 +510,18 @@
if (chan==2)
st_string="stereo";
if (!quiet)
- if (with_vbr)
- fprintf (stderr, "Encoding %d Hz %s audio in %d sample packets at %0.3fkbit/sec (%d maximum bytes per packet) with bitstream version %d\n",
- header.sample_rate, st_string, frame_size, bitrate, bytes_per_packet,bitstream);
+ if (with_cbr)
+ fprintf (stderr, "Encoding %.0f kHz %s audio in %.0fms packets at %0.3fkbit/sec (%d bytes per packet, CBR) with bitstream version %d\n",
+ header.sample_rate/1000., st_string, frame_size/(float)header.sample_rate*1000., bitrate, bytes_per_packet,bitstream);
else
- fprintf (stderr, "Encoding %d Hz %s audio in %d sample packets at %0.3fkbit/sec (%d bytes per packet) with bitstream version %d\n",
- header.sample_rate, st_string, frame_size, bitrate, bytes_per_packet,bitstream);
+ fprintf (stderr, "Encoding %.0f kHz %s audio in %.0fms packets at %0.3fkbit/sec (%d bytes per packet maximum) with bitstream version %d\n",
+ header.sample_rate/1000., st_string, frame_size/(float)header.sample_rate*1000., bitrate, bytes_per_packet,bitstream);
}
/*Initialize CELT encoder*/
st = celt_encoder_create(mode, chan, NULL);
- if (with_vbr)
+ if (!with_cbr)
{int tmp = (bitrate*1000);
if (celt_encoder_ctl(st, CELT_SET_VBR_RATE(tmp)) != CELT_OK)
@@ -731,7 +731,7 @@
bytes_written += ret;
}
- if (with_vbr && !quiet)
+ if (!with_cbr && !quiet)
fprintf (stderr, "Average rate %0.3fkbit/sec, %d peak bytes per packet\n", (total_bytes*8.0/((float)nb_encoded/header.sample_rate))/1000.0, peak_bytes);
celt_encoder_destroy(st);