Chris@0: 
Chris@0:  libmad - MPEG audio decoder library
Chris@0:  Copyright (C) 2000-2004 Underbit Technologies, Inc.
Chris@0: 
Chris@0:  $Id: CHANGES,v 1.14 2004/02/17 02:02:03 rob Exp $
Chris@0: 
Chris@0: ===============================================================================
Chris@0: 
Chris@0: Version 0.15.1 (beta)
Chris@0: 
Chris@0:   * Updated to autoconf 2.59, automake 1.8.2, libtool 1.5.2.
Chris@0: 
Chris@0:   * Replaced Layer III IMDCT routine with one based on a faster algorithm,
Chris@0:     improving both speed and accuracy.
Chris@0: 
Chris@0:   * Improved portability of the Huffman table initialization.
Chris@0: 
Chris@0:   * Fixed a problem that could result in an assertion failure in layer3.c
Chris@0:     due to an invalid Layer III free format bitrate.
Chris@0: 
Chris@0:   * Improved the robustness of Layer II bitrate/mode combinations, and added
Chris@0:     a new MAD_ERROR_BADMODE error enum. The allowability of low-bitrate
Chris@0:     stereo streams is influenced by the --enable-strict-iso option to
Chris@0:     `configure'.
Chris@0: 
Chris@0: Version 0.15.0 (beta)
Chris@0: 
Chris@0:   * Updated to autoconf 2.57, automake 1.7.5, libtool 1.4.3.
Chris@0: 
Chris@0:   * Added new mad_f_div() API routine.
Chris@0: 
Chris@0:   * Added a 64th entry to the Layer I/Layer II scalefactor table, for better
Chris@0:     compatibility with existing streams. The --enable-strict-iso option to
Chris@0:     `configure' can be used to disable use of this entry.
Chris@0: 
Chris@0:   * Modified the header decoding routine to allow the reserved emphasis
Chris@0:     value, for better compatibility with existing streams. The
Chris@0:     --enable-strict-iso option to `configure' can be used to restore the
Chris@0:     previous behavior of reporting this value as an error.
Chris@0: 
Chris@0:   * Added new MAD_EMPHASIS_RESERVED enumeration constant.
Chris@0: 
Chris@0:   * Fixed a bug in the ARM version of mad_f_scale64() discovered by Andre
Chris@0:     McCurdy.
Chris@0: 
Chris@0:   * Rewrote PowerPC assembly for minor gains.
Chris@0: 
Chris@0:   * Modified mad_timer_fraction() to avoid the possibility of division by
Chris@0:     zero when 0 is passed as the second argument.
Chris@0: 
Chris@0:   * Fixed a non-fatal problem caused by attempting to designate ancillary
Chris@0:     bits in Layer III after a decoding error.
Chris@0: 
Chris@0:   * Changed to build a shared library by default.
Chris@0: 
Chris@0:   * Changed to use native Cygwin build by default; give --host=mingw32 to
Chris@0:     `configure' to use MinGW (and avoid a dependency on the Cygwin DLL).
Chris@0: 
Chris@0: Version 0.14.2 (beta)
Chris@0: 
Chris@0:   * Changed Cygwin builds to use MinGW; resulting Win32 executables no
Chris@0:     longer have a dependency on Cygwin DLLs.
Chris@0: 
Chris@0:   * Added a new mad_stream_errorstr() API function to libmad for retrieving
Chris@0:     a string description of the current error condition.
Chris@0: 
Chris@0: Version 0.14.1 (beta)
Chris@0: 
Chris@0:   * Updated config.guess and config.sub to latest upstream versions.
Chris@0: 
Chris@0:   * Enabled libtool versioning rather than release numbering.
Chris@0: 
Chris@0:   * Improved the documentation in minimad.c.
Chris@0: 
Chris@0:   * Several other small fixes.
Chris@0: 
Chris@0: Version 0.14.0 (beta)
Chris@0: 
Chris@0:   * Added a 64-bit FPM negation operation to improve performance of subband
Chris@0:     synthesis on some platforms.
Chris@0: 
Chris@0:   * Improved MSVC++ portability and added MSVC++ project files.
Chris@0: 
Chris@0:   * Added rounding to Layer III requantization for slightly better accuracy.
Chris@0: 
Chris@0: Version 0.13.0 (beta)
Chris@0: 
Chris@0:   * Ancillary data is now properly extracted from Layer III streams.
Chris@0: 
Chris@0:   * Rewrote the Layer III joint stereo decoding routine to correct a major
Chris@0:     MPEG-2 problem and a minor MPEG-1 problem decoding intensity stereo.
Chris@0: 
Chris@0:   * Eliminated the dependency on sign-extending right shifts for Layer I and
Chris@0:     Layer II.
Chris@0: 
Chris@0:   * Renamed `private' field to `private_bits' for better C++ compatibility.
Chris@0: 
Chris@0:   * Gratuitously renamed `sfreq' field to `samplerate' and
Chris@0:     MAD_ERROR_BADSAMPLEFREQ constant to MAD_ERROR_BADSAMPLERATE.
Chris@0: 
Chris@0:   * Added `samplerate' and `channels' fields to synth.pcm struct to allow
Chris@0:     these to be different from the decoded frame, and for simpler access.
Chris@0: 
Chris@0:   * Added new mad_stream_options() and mad_decoder_options() API entries for
Chris@0:     special runtime decoding options.
Chris@0: 
Chris@0:   * Added new MAD_OPTION_IGNORECRC and MAD_OPTION_HALFSAMPLERATE options.
Chris@0: 
Chris@0:   * Added new MAD_FLAG_FREEFORMAT indicator flag.
Chris@0: 
Chris@0:   * Fixed some bugs in the async decoder.
Chris@0: 
Chris@0:   * Added a new mad_timer_multiply() API routine.
Chris@0: 
Chris@0:   * Eliminated `+' from asm constraints under Intel for better compatibility
Chris@0:     with some compilers.
Chris@0: 
Chris@0:   * Fixed a PIC-related problem in imdct_l_arm.S.
Chris@0: 
Chris@0:   * Eliminated a static variable to make libmad thread-safe.
Chris@0: 
Chris@0: Version 0.12.5 (beta)
Chris@0: 
Chris@0:   * Modified Layer III requantization to occur during Huffman decoding for
Chris@0:     significant performance gains.
Chris@0: 
Chris@0:   * Optimized short block IMDCT by eliminating redundant calculations.
Chris@0: 
Chris@0:   * Made several other Layer III performance improvements; added
Chris@0:     ASO_INTERLEAVE1, ASO_INTERLEAVE2, and ASO_ZEROCHECK
Chris@0:     architecture-specific options for best performance on various
Chris@0:     architectures.
Chris@0: 
Chris@0:   * Optimized synthesis DCT to store result values as soon as they are
Chris@0:     calculated.
Chris@0: 
Chris@0: Version 0.12.4 (beta)
Chris@0: 
Chris@0:   * New PowerPC fixed-point assembly courtesy of David Blythe.
Chris@0: 
Chris@0:   * Reorganized fixed-point assembly routines for easier maintenance and
Chris@0:     better performance.
Chris@0: 
Chris@0:   * Improved performance of subband synthesis through better indexing and
Chris@0:     fewer local variables.
Chris@0: 
Chris@0:   * Added alias reduction for the lower two subbands of mixed short blocks,
Chris@0:     per a report of ambiguity with ISO/IEC 11172-3 and for uniformity with
Chris@0:     most other implementations. Also improved alias reduction performance
Chris@0:     using multiply/accumulate.
Chris@0: 
Chris@0:   * Added --enable-strict-iso option to `configure' to override best
Chris@0:     accepted practices such as the alias reduction for mixed short blocks.
Chris@0: 
Chris@0:   * Improved performance of Layer III IMDCT by using longer
Chris@0:     multiply/accumulate runs where possible.
Chris@0: 
Chris@0: Version 0.12.3 (beta)
Chris@0: 
Chris@0:   * Added MPEG 2.5 support.
Chris@0: 
Chris@0:   * Added preliminary support for parameterizing the binary point position
Chris@0:     in the fixed-point representation.
Chris@0: 
Chris@0:   * Added multiply/accumulate optimization to the Layer III IMDCT for long
Chris@0:     blocks.
Chris@0: 
Chris@0:   * Fixed a bug in the handling of Layer III mixed_block_flag.
Chris@0: 
Chris@0:   * Fixed a configure problem when multiple -O CFLAGS are present.
Chris@0: 
Chris@0: Version 0.12.2 (beta)
Chris@0: 
Chris@0:   * Rearranged the synthesis polyphase filterbank memory vector for better
Chris@0:     locality of reference, and rewrote mad_synth_frame() to accommodate,
Chris@0:     resulting in improved performance.
Chris@0: 
Chris@0:   * Discovered a combination of compiler optimization flags that further
Chris@0:     improve performance.
Chris@0: 
Chris@0:   * Changed some array references in layer3.c to pointer derefs.
Chris@0: 
Chris@0: Version 0.12.1 (beta)
Chris@0: 
Chris@0:   * Resolved the intensity + MS joint stereo issue (a simple bug).
Chris@0:     OPT_ISKLUGE is no longer considered to be a kluge.
Chris@0: 
Chris@0:   * Fixed another, hopefully last main_data memory bug.
Chris@0: 
Chris@0:   * Split part of struct mad_frame into struct mad_header for convenience
Chris@0:     and size.
Chris@0: 
Chris@0: Version 0.12.0 (alpha)
Chris@0: 
Chris@0:   * Changed the build environment to use automake and libtool. A libmad
Chris@0:     shared library can now be built using the --enable-shared option to
Chris@0:     `configure'.
Chris@0: 
Chris@0:   * Added another callback to MAD's high-level decoder API after the frame
Chris@0:     header has been read but before the frame's audio data is decoded.
Chris@0: 
Chris@0:   * Streamlined header processing so that mad_frame_decode() can be called
Chris@0:     with or without having already called mad_frame_header().
Chris@0: 
Chris@0:   * Fixed some other header reading miscellany, including CRC handling and
Chris@0:     free bitrate detection, and frame length verification with free
Chris@0:     bitrates.
Chris@0: 
Chris@0:   * Fixed a problem with Layer III free bitrates > 320 kbps. The main_data
Chris@0:     buffer size should now be large enough to handle any size frame, by
Chris@0:     virtue of the maximum possible part2_3_length.
Chris@0: 
Chris@0:   * Further developed the async API; arbitrary messages can now be passed to
Chris@0:     the subsidiary decoding process.
Chris@0: 
Chris@0:   * Streamlined timer.c and extended its interface. It now has support for
Chris@0:     video frame/field lengths, including output support for drop-frame
Chris@0:     encoding.
Chris@0: 
Chris@0:   * Replaced many constant integer preprocessor defines with enums.
Chris@0: 
Chris@0: Version 0.11.4 (beta)
Chris@0: 
Chris@0:   * Fixed free format bitrate discovery.
Chris@0: 
Chris@0:   * Changed the timer implementation and extended its interface.
Chris@0: 
Chris@0:   * Integrated Nicolas Pitre's patch for pre-shifting at compile-time and
Chris@0:     for better multiply/accumulate code output.
Chris@0: 
Chris@0:   * Applied Simon Burge's patch to imdct_l_arm.S for a.out compatibility.
Chris@0: 
Chris@0:   * Added -mtune=strongarm for all ARM targets.
Chris@0: 
Chris@0: Version 0.11.3 (beta)
Chris@0: 
Chris@0:   * Added new --enable-speed and --enable-accuracy options for `configure'
Chris@0:     to automatically select appropriate SSO/ASO options, et al.
Chris@0: 
Chris@0:   * Modified subband synthesis to use multiply/accumulate optimization (if
Chris@0:     available) for better speed and/or accuracy.
Chris@0: 
Chris@0:   * Incorporated Andre McCurdy's changes for further rounding optimizations
Chris@0:     in the rest of his code.
Chris@0: 
Chris@0: Version 0.11.2 (beta)
Chris@0: 
Chris@0:   * Incorporated Nicolas Pitre's ARM assembly and parameterized scaling
Chris@0:     changes.
Chris@0: 
Chris@0:   * Incorporated Andre McCurdy's ARM assembly optimization (used only if
Chris@0:     --enable-aso is given to `configure' to enable architecture-specific
Chris@0:     optimizations.)
Chris@0: 
Chris@0:   * Reduced FPM_INTEL assembly to two instructions.
Chris@0: 
Chris@0:   * Fixed accuracy problems with certain FPM modes in synth.c.
Chris@0: 
Chris@0:   * Improved the accuracy of FPM_APPROX.
Chris@0: 
Chris@0:   * Improved the accuracy of SSO.
Chris@0: 
Chris@0:   * Improved sync discovery by checking for a sync word in the following
Chris@0:     frame.
Chris@0: 
Chris@0:   * Minor code clean-up.
Chris@0: 
Chris@0:   * Added experimental rules for generating a libmad.so shared library.
Chris@0: 
Chris@0: Version 0.11.1 (beta)
Chris@0: 
Chris@0:   * Moved libmad code into a separate directory.
Chris@0: 
Chris@0:   * Changed SSO to be disabled by default, as output accuracy is deemed to
Chris@0:     be more important than speed in the general case.
Chris@0: 
Chris@0:   * Fixed a bug in Layer III sanity checking that could cause a crash on
Chris@0:     certain random data input.
Chris@0: 
Chris@0:   * Extended the Layer III requantization table from 8191 to 8206 as some
Chris@0:     encoders are known to use these values, even though ISO/IEC 11172-3
Chris@0:     suggests the maximum should be 8191.
Chris@0: 
Chris@0: Version 0.11.0 (beta)
Chris@0: 
Chris@0:   * Implemented MPEG-2 extension to Lower Sampling Frequencies.
Chris@0: 
Chris@0:   * Improved Layer III performance by avoiding IMDCT calculation when all
Chris@0:     input samples are zero.
Chris@0: 
Chris@0:   * Significantly reduced size of Layer II tables.
Chris@0: 
Chris@0: Version 0.10.3 (beta)
Chris@0: 
Chris@0:   * Improved SSO output quality.
Chris@0: 
Chris@0:   * Made portable to cygwin.
Chris@0: 
Chris@0:   * Localized memory references in III_huffdecode() for better performance.
Chris@0: 
Chris@0: Version 0.10.2 (beta)
Chris@0: 
Chris@0:   * Rewrote Layer III long block 36-point IMDCT routine for better
Chris@0:     performance.
Chris@0: 
Chris@0:   * Improved subband synthesis fixed-point games somewhat.
Chris@0: 
Chris@0: Version 0.10.1 (beta)
Chris@0: 
Chris@0:   * Added a subband synthesis optimization (SSO) which involves modifying
Chris@0:     the fixed-point multiplication method during windowing. This produces
Chris@0:     subtle differences in the output but improves performance greatly.
Chris@0: 
Chris@0:   * Added I_STEREO and MS_STEREO flags to frame struct.
Chris@0: 
Chris@0:   * Eliminated privately-used CRCFAILED flag.
Chris@0: 
Chris@0:   * Fixed a bug where Layer III decoding could crash on some badly-formatted
Chris@0:     (e.g. non-MPEG) bitstreams.
Chris@0: 
Chris@0:   * Miscellaneous code clean-up.
Chris@0: 
Chris@0: Version 0.10.0 (beta)
Chris@0: 
Chris@0:   * Added SPARC fixed-point math support.
Chris@0: 
Chris@0:   * Revamped libmad API for better high- and low-level support.
Chris@0: 
Chris@0:   * Documented more of the code.
Chris@0: 
Chris@0:   * Changed sync semantics such that new stream buffers are assumed to be
Chris@0:     sync-aligned.
Chris@0: 
Chris@0:   * Changed Layer III to dynamically allocate static memory so as not to
Chris@0:     waste it (about 6.4K) when only decoding Layer I or Layer II.
Chris@0: 
Chris@0: ===============================================================================
Chris@0: