v0.1a - initial release (7 Nov 1998) v0.1b - released 1 Oct 99 - filter subband and window subband from Lame3.x added - cleaned up some memory stuff in psy.c - a fast mode switch (-f) which turns off psychoacoustics -instead of having the psymodel calculate SNR just use a bunch of preset values. I just made up some SNR values based upon my test file. These could probably be tuned so that there's a set of SNR values for rock, pop, classical, voice etc. v0.1d - released 14 Oct 99 - cleaned up some code - removed decoding source code (use mpg123 instead :) - renamed subs.c to fft.c (which is mostly what it does :) - moved the filter/window_subband into subband.c - split encode.c into generic code (encode), LayerI code (encodeI) and layerII code (encodeII) - no longer use obtain_parameters(). interactive input is a pain v0.1e - released 15 Oct 99 - cleaned up some code - defaults to 192kbps, psy model 1 with extension mp2 (see encoder.h) - added (-a) downmix to mono option (code in encode.c/get_audio() ) - transplanted FHT-based fft() from LAME. - II_f_f_t() now calls fht(). heaps faster. (in tonal.c used by psymodel 1) This could be neater though. - PDS's II_scale_factor_calc() put in encodeII.c v0.1f - released 16 Oct 99 - PDS's updated scale_factor_calc - integrated a few functions in tonal.c/II_psycho_one() which gave about a 15% speedup when using psymodel 1 v0.1g - released - tweaked tonal.c/threshold() 10-15% improvement in this routine. v0.2a - released 24 october 1999 - changed base source code from mpegaudio.tar to the dist10/lsf/encoder. This new code includes lower sampling frequency (LSF) extensions. - added in most of the changes already done. - LSF mode is faulty. Seg faults after about 190 frames here. LSF mode works if you turn off psy models though (-f) v0.2b - released 30 October 1999 - added a "-q n" mode. will only calculate the psy model every n frames. "-q 0" is equivalent to "-f". - reimplemented "downmix to mono" -a switch - WAV file support from Nick Burch - some code cleanup - removed more LayerI & III stuff. v0.2c - skipped this release v0.2d - released 6 November 1999 - used Mathew Hendry's enhanced fht() with table lookup (20% faster) - made add_db an inline function and tweaked it (25% faster) - code cleanup - renamed psy.c->psycho_II.c, tonal.c->psycho_I.c, other misc stuff - added a "no padding" (-r) switch. Needed by the rfa.org (thanks to Federico Grau) - tweaked threshold 33% speedup. v0.2e - released 14 December 1999 - Some minor speedups from PDS in encode.c - Changed // comments to /* */ style - Merged some functions in subband.c - Cleaned up file input parsing a little. - Now reads PCM from /dev/stdin if this is specified as input filename (kludgy but works for me) - Started putting tables into exe. (only enwindow done so far) v0.2f - released 15 April 2000 - mainly cosmetic changes in the source and code layout - removed a lot of unused functions, and split common.c into more logical subumits. - about 15-20% faster when using psymodel I. (merged some functions in psycho_i.c) v0.2g - released 22 April 2000 - added a VBR mode. See README.VBR - cleaned up more code. v0.2h - stdio/stdout now parse a little better using '-' - tables now included in the exe thanks to Oliver Lietz (I haven't had a chance to test it much though) - including the tables in the exe seems to have boosted speed quite a bit (25%?) - re-implemented byte-swapping in get_audio.c (command line switch -x ) v0.2i - Released 13 July 2001 - DAB extensions from Phillipe Jouguet - Added a switch to swap the input channels (can someone confirm that toolame swaps them incorrectly by default) v0.2j - Released 13 February 2003 - Definitely LGPL - VBR mode is stable for all sampling frequencies. (See README.VBR for details) - speedups and cleanups from Phillipe Jouguet - philippe.jouguet at vdl.fr (spelling, LSF using psyII, WAVE reading) Henrik Herranen - leopold at vlsi.fi (WAVE reading) Andreas Neukoetter - anti at webhome.de (verbosity patch '-t' switch for transcode plugin) Sami Sallinen - sami.sallinen at g-cluster.com filter_subband loop unroll psycho_i fix for "% 1408" calcs) v0.2k - Released 16 February 2003 - Nicolas Croiset unlimited file read size for stdin fixed bitstream ending so that mp2 files can be concatenated DAB options now comes with a size argument (finally?) fixed the core dump in psycho_1 - Ricardo Schelp ricardoschelp at arnet.com.ar combined window/filter subband routine. nice speed increase. v0.2l - Released 3 March 2003 - Added new psychomodels -1 use precomputed values 0 very low complexity 1 same old psy model 1 2 same old psy model 2 3 psy model 1 rewritten from scratch 4 rewritten psy model 2 - Rewrote the bitstream encoding routines - Sped up the sin/cos/atan stuff in psymodel 4 (using tables now) v0.2l with ZMQ output - February 2014 - added outputzmq.{h, c} Matthias P. Braendli mike (mikecheng at NOT planckenergy.com) [remove the NOT]