ref: 1c10295d291b041e9466949b47d3e4dde39d8380
dir: /ChangeLog/
2001-05-31 menno Bugfix for linux 2001-05-30 menno Changed license from GPL to LPGL Higher release version Fixed linux compilation 2001-05-29 menno Fixed bit counting, file length calculation works again 2001-05-28 menno Small bugfix in backward predictor Changes for ISO MPEG4 compatibility (sorry :(( ) Old MPEG4 files with ADTS header will not be playable anymore 2001-05-18 menno Cleaned up backward prediction code Added encoder library interface documentation 2001-05-11 oxygene2000 - new makefile for unix-like systems (should also work with cygwin) based on autoconf/-make and libtool 2001-05-09 menno Fixed NSIS install script to avoid warnings Added libsndfile to the CVS tree, making compilation a lot easier 2001-05-08 menno Better support for main profile in frontends 2001-05-02 menno Added support for MAIN AAC object type 2001-04-29 flyingfox - disabled job list persistence (file format incompatibility) - disabled job list multiple selection (property merging is not working properly anyway) - added MPEG version support - changed AAC profile to AAC object type - disabled LTP under MPEG version 2 (if LTP selected it will go back to LC) 2001-04-27 flyingfox fixed debug build fixed compile error 2001-04-19 menno Changes in MPEG4 file format, mainly comments for clarification 2001-04-11 menno Fixed MPEG4 object type problem 2001-04-10 menno Fixed another MPEG4/2 issue TNS order Big update of GUI 2001-04-03 menno ADTS layer fixed for MPEG-4 Finally correct MPEG4 ADTS headers Re: ADTS layer fixed for MPEG-4 2001-04-01 xfhobbes added general user input tracking during job list processing and question for target directory creation where appropriate improved progress counter; added general user input tracking during job list processing; added column in the job list ctrl to display the outcome of job processing 2001-03-19 menno Table data fix for 8kHz Correct encoding of files other than 2 channel now 2001-03-18 xfhobbes Major update: roughly consolidated filter job functionality, added shell open support and several other things, cleaned up scc files and ignore list Moved beep after completed processing from individual beep for every job to a global one per processing command. Rewrote DoProcessing() method to comply to rules of Hungarian Notation and to common C++ code styles. 2001-03-18 menno Changed NSIS installation script now it supports opening of .jbl (joblist) files automatically Added NSIS (www.nullsoft.com) installation script 2001-03-17 menno Added MPEG2 AAC support (compile with MPEG2AAC defined) make sure you don't use LTP when using MPEG2 AAC, because it is not supported in MPEG2 AAC 2001-03-15 menno Small bugfix 2001-03-14 menno New GUI application 2001-03-13 menno Updated .cvsignore 2001-03-12 menno cleanup Small change in API 2001-03-06 menno Fix in reconstruction of spectrum before LTP buffer update -added profile selection -some cleanup 2001-03-05 menno Added LTP Needs some more refinement, but it works LTP speedup 2001-03-05 oxygene Updated for LTP Included math.h for math functions in linux. 2001-03-02 menno unix fix 2001-03-01 menno unix fix 2001-03-01 oxygene Added support for TNS, fixed a typo Install and uninstall target added. Small change 2001-02-28 menno - added TNS support - some other cleanups Support for TNS in frontends 2001-02-26 oxygene Added min() function for unix Added min() and max() functions for unix Included math.h to make unix compiler happy Added min() function for unix. 2001-02-25 menno Added calculation of KBD window 2001-02-12 menno Different MS switching model 2001-02-10 menno - Fixed scalefactor bandwidths (Thanks again to Ivan) - Better checking of configuration 2001-02-09 menno Scalefactor bandwidth update (Thanks to Ivan) 2001-02-08 menno Fixed scalefactorband data (thanks to Ivan) 2001-02-05 menno changed LINUX -> __unix__ 2001-02-04 oxygene2000 - new Makefiles - .cvsignore for a cleaner cvs-output - // xyz -> /* xyz */ for ansi-c compliance - minor fixes to work on linux 2001-02-01 menno Fixed memory leakage 2001-01-31 menno Completely new psychoacoustic model 2001-01-25 menno Encoding now done in a thread. Plus some other small changes. Small changes 2001-01-24 menno Minor changes to GUI 2001-01-23 menno Fix to remove compiler warnings Added graphical user interface ADTS headers supported 2001-01-19 menno Changed ATH formula Speedup in huffman coder 2001-01-17 menno Added frequency cut off filter 2000-12-04 menno Changed usage message a bit 2000-11-10 menno Bugfix in psychoacoustics and window shape switching changed 2000-11-07 menno Quantizer update. No more quantizer crashes 2000-11-01 menno Code reorganisation 2000-10-31 menno Cleanup: more sensible organisation 2000-10-08 menno Bugfixed and commented psychoacoustic module Removed unused file 2000-10-06 menno Code cleanup 2000-10-05 menno Big code cleanup Copyright message cleanup. CVS keywords added 2000-10-04 menno Wrong bits per frame got calculated LFE is now also working correctly 2000-10-03 menno Multichannel fixes. All working now, except LFE channels 2000-10-01 menno Small fix in LFE frequency cut-off Encoder LFE fixup. Should be working now, but no way to test it yet. 2000-09-30 menno Some variable names changed (purely cosmetic) 2000-09-28 menno Release updates 2000-09-25 menno Small changes Added VBR quantizer (but not working yet) 2000-08-24 menno Cleanup and small quality update Fixed a bug in MS threshold calculation 2000-08-23 menno Different bark formula 2000-08-20 menno Fixed Makefile 2000-08-11 menno Bugfixes and multichannel updates removed some redundant code 2000-08-10 menno Cleanup bugfix removed IS 2000-08-08 menno Removed IS. It is useless Changed for multichannel support. And big cleanup Changed for multichannel support 2000-07-22 prkoat Support for SCE (mono files) 2000-05-18 menno Tuned short block switching 2000-05-16 menno First go at multichannel support (not working at all, yet) 2000-04-05 oxygene2000 - Cosmetic change (gcc complains about missing newline at EOF) - Made it compile with gcc/linux again With -DLINUX faac shows the cpu-time needed to encode not real time, therefore better measurement is possible 2000-03-29 menno Quality update 2000-03-28 menno Separated quantization and noiseless coding Made TNS default again Moved a function Small change 2000-03-23 lenox restored gain control bits in the stream 2000-03-22 lenox added new parameters for 'block' mode 2000-03-17 lenox removed function faac_EncodeFinish() 2000-03-13 lenox added support for block input (for DLL interface without input files) 2000-03-12 lenox removed gain control & changed win32 defines 2000-03-02 lenox new quantizer code for other than VC and BCB compilers added gcc asm code 2000-02-28 lenox huge cleanup (for unused parameters and vars) slightly changed arg parsing improved readability of the code removed unused parameters added BC asm compatibility 2000-02-28 menno Bugfix 2000-02-27 menno Changed short block quantization 2000-02-24 menno Now using LAME's mid/side threshold calculation 2000-02-24 thebard Removed remainign thread and semephore code After more compilation, there is not any function thats taking more than 10% of the total time (see the profile graph) For fun I added threads to psy6, but since its only 7 seconds, the constant rebuilds of the threads gain little improvement, and will suck on non multiple cpu boxes, so I removed it. 2000-02-23 lenox fixed wildcards processing 2000-02-23 thebard Now encoding faster than realtime with suggested compiler options in Makefile. Still alot of room left for assembler and or thread tweaking. Alterations so that we can thread functions for multiple CPU's Got run time down 100%. It looks like gcc is starving registers or causing mosterous cache misses. More to come. 2000-02-23 menno Quality update Removed winswitch.c 2000-02-22 thebard Added the profiling text file. inlined output_bits, not that it matters much in performance... Macroized double_to_int. Shaved off a few seconds (avoided all the function call overhead) Makefile changes for default compile and nots on options, Also added an include so that faac.c links. Changes the ending return to an exit so we can profile. Timer now works correctly for unix. 2000-02-22 menno Made ADTS quality same as other files quality 2000-02-19 lenox restructure added faac.c 2000-02-19 oxygene Uninstall target added 2000-02-18 lenox cleanup new interface functions 2000-02-18 menno Added libsndfile.lib 2000-02-17 lenox added support for all headers 2000-02-17 menno Removed some debug code Bugfixes for bit counting 2000-02-16 oxygene Install target added. 2000-02-15 lenox restructure (aacconfig eliminated) new parsing function 2000-02-15 menno Made DLL compile Fixed DLL Added ADTS header writing 2000-02-14 lenox new rft (reference to fftw removed, sorry :) ) removed rdft due to new real fft removed rdft stuff 2000-02-09 lenox fix to use without fttw 2000-02-09 oxygene2000 Added additional implementation of rdft using fftw (from rank 1 to rank >20 in my profiler = a lot faster) Added fixed fftw-stuff. Use rdft_spectrum2.c instead of rdft_spectrum.c and set -DUSE_FFTW in Makefile to enable it.. Be sure that you have libfftw installed fftw_init() and fftw_destroy() have to be placed elsewhere (should only be run once in the program -> speed) Another mistake, argh! Placed fftw_init/destroy where they have to be Removed transfo.h as it isn't needed Added rdft.h 2000-02-09 menno Restored small mistakes. 2000-02-08 lenox removed old FFT & some restucture 2000-02-08 menno Made advantage of Oxygene2000's optimization, to use more accurate delay estimation Better quantization of short blocks 2000-02-08 oxygene2000 really small speed-up in psy_step6 Why calculating sqrt(2) if it is present as constant? 2000-02-07 oxygene2000 Another speedup by solving calculations into something more simple but equivalent Fixed missing dependency rateconv.c Speeded up estimate_delay Commented out unused loop Made more variables to tune gcc 2000-02-07 lenox some cleanup cleanup added support for different window shapes bugfix fixed bug with atan2() added kbd window tables 2000-02-07 menno Changed last changes back, they made no difference Bug fix Change in window switching parameters Changed command line (TNS is now off by default) One step closer to window shape switching Bugs fixed in psych Same change to short blocks as oxy2000 made to long blocks 2000-02-06 menno Quite some bug fixes Speedup 2000-02-05 menno Psych changed and pulse coding updated SMR values were one frame ahead of data. Big quality improvement. Bugfixes Fixed lower bitrates (continued) More fixes Fixed lower bitrate encoding 2000-02-04 menno Psycho speedup Fixed mistake. This caused a very bad sound Bugs fixed Window switching back to old 2000-02-03 menno Printing options to screen bugfix for blocks with only 0's 2000-02-02 menno Bug fixes Serious bugfixes and speedups Big quantizer change 2000-01-31 lenox cleanup bugfix & cleanup 2000-01-31 menno Reduced psycho complexity (same result) 2000-01-30 menno FFT change (cleanup) 2000-01-29 lenox removed imdct.c changed T/F transformation 2000-01-27 menno New block switching (sounds bad for now) 2000-01-24 lenox removed unneeded stuff for LTP off 2000-01-24 menno Unneeded calls to functions removed Command line parsing fixed 2000-01-21 menno Updated readme Bug fix Version upgraded 2000-01-19 menno LTP fixed and some other bugs. 2000-01-18 menno Fixed encoding of some samplerates Oops Added frequency cut-off option (-c) 2000-01-17 menno Added sampling rate conversion (-s), buggy with some sample rates 2000-01-14 menno First attempt at doing PNS (use -pns). Also every bitrate higher than 16kbps is now supported in step of 1kbps 2000-01-12 menno Cleanup 2000-01-10 menno Small fix Added support for all sampling rates All psycho values now calculated by formulas Full support of all sample rates, and some other fixes 2000-01-09 lenox new dirs for output 2000-01-09 menno Bugfix restored previous change 2000-01-08 lenox some cleanup 2000-01-07 lenox fixed bug in block switching (was since 1.16) 2000-01-07 menno Added option to disable LTP (-nl) 2000-01-06 menno Cleanup Updated delay detection in LTP Added support for PNS (but it is not used at the moment) Small quality fixes Divide by zero bugfix Mistake fixed Fix 2000-01-06 lenox new restructured mdct and fft 2000-01-05 menno Added Long Term Prediction (LTP), new decoder is needed for this Changed block switching Removed backward prediction 2000-01-03 lenox Added 2048 FFT 1999-12-30 menno Structure cleanup 1999-12-29 menno Restored old FFT due to some serious bugs 1999-12-23 menno New fft in psycho model (fast) and new unpredictabilty measure calculation Quick bugfix after latest update Quality fix, in MS protection ratio calculation Removed the need for 4 channel buffers. Now there are 2 Little fix Fix in Mid/Side calculation Added fft.c 1999-12-22 menno Changed the DLL interface version. Streamlined spreading function. Big speedup 1999-12-20 menno Ooops, quick bugfix after latest changes gave an extra bug Very big speedup Changed TNS switching 1999-12-20 lenox some cleanup 1999-12-19 menno Bugfix preventing memory overflow Bugfix: MS first gave some distortion in the right channel. Because the Side channel was incorrectly formed. Changed InputBuffer into double instead of float (less conversions) Added some extra options 1999-12-19 lenox new FFT code added fastfft.c new FFT code & converted vars to double removed reference to transfo.h 1999-12-17 lenox removed util.c 1999-12-16 menno New and cleaner energy calculation in ms.c Serious bugfix in psych.c and removed all global statics, they are just normal global variables now. Code cleanup and one bugfix Fixed bug with bitBuffer and better ADIF header writing Fixed bug with calculating number of bytes from the number of bits. Put MSEnergy() after MSPreprocess(). Off course. 1999-12-15 lenox Removed util.c some speed optimization and restructure 1999-12-13 menno some bugfixes