Newer
Older
monitord / lame-3.97 / doc / html / .svn / text-base / history.html.svn-base
@root root on 23 Jan 2012 76 KB Migration from SVN revision 455
<!DOCTYPE html PUBLIC "-//w3c//dtd html 4.0 transitional//en">
<html>
<head>
<meta http-equiv="Content-Type"
 content="text/html; charset=iso-8859-1">
<meta name="GENERATOR"
 content="Mozilla/4.7 [en] (X11; I; Linux 2.2.12-20 i686) [Netscape]">
<meta http-equiv="content-type"
 content="text/html; charset=ISO-8859-1">
<title>LAME Changelog</title>
</head>
<body text="#000000" bgcolor="#ffffff" link="#888888" vlink="#555555"
 alink="#bbbbbb">
<center>
  <h1> History</h1>
</center>
Starting with LAME 3.0: <br>
<font color="#ff0000">red = features and bug fixes which affect quality</font> 
<br>
<font color="#3366ff">blue = features and bug fixes which affect speed</font> 
<br>
black = usability, portability, other 
<hr>
<h3>LAME 3.97&nbsp; &nbsp;September 24 2006</h3>
<ul>
  <li>3.97 beta 3 becomes 3.97 
  </li>
</ul>
<br>
<h3>LAME 3.97 beta 3&nbsp; &nbsp;August 19 2006</h3>
<ul>
  <li>Gabriel Bouvigne: 
    <ul>
      <li><font color="#ff0000">Workaround against a short blocks detection issue</font></li>
    </ul>
  </li>
</ul>
<br>
<h3>LAME 3.97 beta 2&nbsp; &nbsp;November 26 2005</h3>
<ul>
  <li>Gabriel Bouvigne: 
    <ul>
      <li>Fixed an initialization error when input is not using a standard sampling frequency</li>
      <li>Fixed a possible assertion failure in very low bitrate encoding</li>
      <li><font color="#ff0000">Slight change regarding ATH adjustment with V5</font></li>
      <li><font color="#ff0000">Reinstated bit reservoir for 320kbps CBR</font></li>
      <li><font color="#3366ff">ReplayGain analysis should now be faster when encountering silent parts</font></li>
    </ul>
  </li>
  <li>Takehiro Tominaga: 
    <ul>
      <li>Fixed a possible link problem of assembly code</li>
    </ul>
  </li>
</ul>
<br>
<h3>LAME 3.97 beta 1&nbsp; &nbsp;September 12 2005</h3>
<ul>
  <li>Takehiro Tominaga: 
    <ul>
      <li>Fixed an out of array access in mp3rtp
      <li><font color="#ff0000">Fixed a quality setting in DLL</font></li>
      <li>Fixed display when using --silent</li>
    </ul>
  </li>
  <li>Vitaly Ivanov: 
    <ul>
      <li>Updated DirectShow interface</li>
    </ul>
  </li>
  <li>Robert Hegemann: 
    <ul>
      <li><font color="#ff0000">Fixed an out of array access</font></li>
      <li><font color="#ff0000">Fixed some small rounding problem in vbr-new quantization routines</font></li>
      <li><font color="#ff0000">Fixed a bug in vbr-new regarding high frequencies (sfb21) when using -Y</font></li>
      <li><font color="#ff0000">Fixed a few bugs in vbr-new when using -Y</font></li>
      <li><font color="#ff0000">Updated scalefactors allocation scheme in vbr-new</font></li>
      <li>Fixed mingw32 configure problems</li>
      <li>Resolved some compiler warnings</li>
      <li>Updated command-line visualisation</li>
    </ul>
  </li>
  <li>Gabriel Bouvigne: 
    <ul>
      <li>Changed some FLOAT8 to FLOAT</li>
      <li>Added project files for VC7</li>
      <li><font color="#ff0000">Reworked -q1 and -q0</font></li>
      <li><font color="#ff0000">Updated presets</font></li>
      <li><font color="#ff0000">Fixed an error in ISO quantization on systems not using the IEEE754 hack</font></li>
      <li><font color="#3366ff">Faster quantization</font></li>
      <li><font color="#3366ff">SSE version of init_xrpow</font></li>
    </ul>
  </li>
  <li>Rogério Brito: 
    <ul>
      <li>Updated Debian packaging</li>
      <li>Documentation work</li>
    </ul>
  </li>
  <li>Chris Miller: 
    <ul>
      <li>Support for x64 platform SDK in makefile.msvc</li>
    </ul>
  </li>
</ul>
<br>
<h3>LAME 3.96.1 &nbsp; &nbsp;July 25 2004</h3>
<ul>
  <li>Robert Hegemann: 
    <ul>
      <li><font color="#ff0000">Fixed a rare bug in vbr-new (could lead to crashes or data corruption)</font></li>
    </ul>
  </li>
  <li>Gabriel Bouvigne: 
    <ul>
      <li>some fixes in ACM codec</li>
      <li>fixed padding when encoding to 320kbps</li>
      <li><font color="#ff0000">fixed block size selection for mid and side channels</font></li>
    </ul>
  </li>
</ul>
<br>
<h3>LAME 3.96 &nbsp; &nbsp;April 11 2004</h3>
<ul>
  <li>Gabriel Bouvigne:
  <ul>
    <li><font color="#ff0000">new quantization selection mode (used in ABR/CBR)</font></li>
    <li><font color="#ff0000">set sfscale for ABR/CBR up to 160kbps</font></li>
</ul>
</li>
</ul>
<br>
<h3>LAME 3.96 beta 2 &nbsp; &nbsp;March 28 2004</h3>
<ul>
  <li>Takehiro Tominaga:
  <ul>
    <li><font color="#3366ff">removed unnecessary integer convertion in resampling</font></li>
</ul></li>
  <li>Robert Hegemann:
  <ul>
    <li><font color="#ff0000">reworked scalefactor allocation in vbr-new</font></li>
    <li>fixed a freeformat decoding problem</li>
</ul>
  <li>Gabriel Bouvigne:
  <ul>
    <li><font color="#ff0000">updated minimal bitrate for V1 and V2</font></li>
</ul>
</li>
  <li>Aleksander Korzynski:
  <ul>
    <li>added ability to disable ReplayGain analysis</li>
</ul>
</li>
</ul>
<br>
<h3>LAME 3.96 beta &nbsp; &nbsp;March 7 2004</h3>
<ul>
  <li>Takehiro Tominaga:
  <ul>
    <li>fixed decoding issue</li>
</ul></li>
  <li>Aleksander Korzynski:
  <ul>
    <li>changed internal ReplayGain handling</li>
    <li>fixed some issues when ReplayGain is used with resampling</li>
</ul>
</li>
  <li>Robert Hegemann:
  <ul>
    <li>added standard ISO quantization for vbr-new, used at lower quality settings</li>
    <li><font color="#3366ff">faster count_bits for vbr-new</font></li>
    <li><font color="#3366ff">faster find_scalefac_ave function for vbr-new</font></li>
    <li><font color="#ff0000">fixed an out of array access in psychoacoustic models;
        this bug could make some psy calculations worthless and sometimes let lame crash</font></li>
    <li><font color="#ff0000">fixed an error on silent scalefactor bands;
        this bug resulted in huffman data overrun problems while decoding,
        resulting in audible glitches</font></li>
    <li>fixed a freeformat decoding bug</li> 
</ul>
  <li>Gabriel Bouvigne:
  <ul>
    <li><font color="#ff0000">adjusted short block thresholds</font></li>
    <li>fixed some array addressing bugs</li>
    <li>made ReplayGain analysis reentrant</li>
</ul>
</li>
  <li>David Chandler: fixed a crash in quantize_xrpow
</li>
  <li>Michal Bacik: fixed a crash when using 8kHz
</li>
  <li>Goran Markovic: fixed some decoding bugs
</li>
  <li>John Edwards: fixed a too small buffer in ReplayGain code</li>
</ul>
<br>
<h3>LAME 3.95.1 &nbsp; &nbsp;January 12 2004</h3>
<ul>
  <li>Gabriel Bouvigne:
  <ul>
    <li>fixed a crash when using vbr-new</li>
    <li>changed ReplayGain reference level to 89dB</li>
</ul></li>
</ul>
<br>
<h3>LAME 3.95 &nbsp; &nbsp;January 11 2004</h3>
<ul>
  <li>Gabriel Bouvigne:
  <ul>
    <li><font color="#ff0000">fixed lowpass values when using vbr with mono files</font></li>
    <li><font color="#3366ff">faster quantization loops</font></li>
    <li><font color="#3366ff">faster count_bits</font></li>
    <li>fixed a buffer requirement error in ACM codec</li>
  </ul></li>
  <li>Takehiro TOMINAGA:
  <ul>
    <li>fixed mpglib and other decoding support code to prevent the crash when invalid mp3 input</li>
  </ul></li>
  <li>removed Layer I decoding support</li>
  <li><font color="#3366ff">use FastLog and IEEE 754 hack on PowerPC too (approx. 10 percent faster)</font></li>
</ul>
<br>
<h3>LAME 3.94 beta December 15 2003</h3>
<ul>
  <li>Takehiro Tominaga: 
    <ul>
      <li><font color="#ff0000">fixed block switching of nspsytune</font></li>
      <li><font color="#ff0000">best huffman divide in the inner loop.</font> 
        This should improve the quality, but PAINFULLY slow. So it is not enabled 
        by default. Use -q0 to use it.</li>
      <li>Changed -q option mapping. "-q2" until version 3.93 is now "-q3".</li>
      <li><font color="#ff0000">saving bits by better scalefactor storing</font></li>
      <li>removed Vorbis support</li>
      <li><font color="#ff0000">substep quantization.</font>This should help breaking 
        the SFB21 bloating problem</li>
      <li><font color="#ff0000">made psychoacoustic model aware of ATH adjustements</font></li>
      <li><font color="#ff0000">use ATH value as short block masking lower limit</font></li>
      <li><font color="#ff0000">several fixes in psychoacoustic model</font></li>
      <li>more robust decoding</li>
    </ul></li>
  <li>Mark Taylor / Gabriel Bouvigne: fixed issues in VBR header</li>
  <li>Mark Taylor: workaround against some hardware decoder defficiencies</li>
  <li>Aleksander Korzynski: ability to compute the "Radio" ReplayGain and detect clipping 
    on the fly. The ReplayGain value is stored in the Lame tag.</li>
  <li>Gabriel Bouvigne:
  <ul>
    <li><font color="#ff0000">work on presets</font></li>
    <li><font color="#ff0000">use presets by default for cbr/abr</font></li>
    <li><font color="#ff0000">use presets by default for vbr</font></li>
    <li><font color="#ff0000">analog silence detection in partitionned sfb21</font></li>
    <li><font color="#3366ff">do not compute noise in upper 0 part of the spectrum</font></li>
    <li><font color="#3366ff">only compute noise in modified scalefactor bands</font></li>
  </ul></li>
  <li>Guillaume Lessard:
  <ul>
    <li>nogap related changes</li>
  </ul></li>
  <li>Alexander Leidinger:
  <ul>
    <li>prevent closing the input fd prematurely if the input is a named pipe</li>
  </ul></li>
</ul>
<br>
<h3>LAME 3.93.1 &nbsp; &nbsp;December 1 2002</h3>
<ul>
  <li>Gabriel Bouvigne: 
    <ul>
      <li>preset medium added to the dll interface</li>
      <li><font color="#ff0000">fix for abr/cbr presets</font></li>
      <li><font color="#ff0000">fix -q0 switch</font></li>
    </ul>
  </li>
  <li>Alexander Leidinger: fix link problem on systems where socket() resides 
    in libsocket</li>
</ul>
<br>
<h3>LAME 3.93 &nbsp; &nbsp;November 16 2002</h3>
<ul>
  <li>Takehiro Tominaga: 
    <ul>
      <li><font color="#ff0000">bit allocation for pre-echo control improved for 
        single channel encodings</font></li>
      <li><font color="#ff0000">substep noise shaping</font></li>
      <li><font color="#3366ff">optimizations by changing data structure</font></li>
      <li><font color="#ff0000">noise shaping model 2 fix</font></li>
      <li><font color="#3366ff">nspsytune FIR filter clean up</font></li>
      <li><font color="#ff0000">fix small psymodel bugs(DC current estimation, 
        preecho detection of non-VBR mode, and nspsymode initialization)</font></li>
      <li>portability fixes for Tru64 UNIX</li>
    </ul>
  </li>
  <li>Albert Faber: some fixes in the DLL</li>
  <li>Simon Blandford: fixes for channel scaling in mono mode</li>
  <li><font color="#3366ff">Dominique Duvivier: some optimizations and a faster 
    log10 function</font></li>
  <li>Mark Taylor: 
    <ul>
      <li>some tag related fixes in the direct show filter and in the ACM codec</li>
      <li><font color="#3366ff">fixed a mono encoding bug found by Justin Schoeman</font></li>
      <li>calc_noise bug fix</li>
      <li>other fixes</li>
    </ul>
  </li>
  <li>Alexander Leidinger: 
    <ul>
      <li>update to autoconf 2.53, rewrite some configure tests</li>
      <li>Akos Maroy: determine gcc version even with gcc 3.1</li>
      <li>Andrew Bachmann: compile shared libs on BeOS (and perhaps other arches)</li>
      <li>ultrasparc switches for gcc 3.1</li>
      <li>fixes for SunOS 4.x</li>
      <li>fixes for 64bit arches</li>
      <li>CFLAGS fix for IRIX</li>
      <li>don't override CFLAGS if exptopt isn't requested</li>
    </ul>
  </li>
  <li>Robert Hegeman: 
    <ul>
      <li><font color="#3366ff">some fixes</font></li>
      <li><font color="#ff0000">some fixes for VBR</font></li>
    </ul>
  </li>
  <li>Gabriel Bouvigne: 
    <ul>
      <li>--noasm switch. Might help Cyrix/Via users</li>
      <li><font color="#ff0000">presets and alt-presets merged</font></li>
    </ul>
  </li>
</ul>
<br>
<h3>LAME 3.92 &nbsp; &nbsp;April 14 2002</h3>
<ul>
  <li><font color="#ff0000">Alexander Leidinger:&nbsp; add non linear psymodel 
    (compile time option, disabled by default)</font>, workaround a bug in gcc 
    3.0.3 (compiler options, based upon suggestions from various people, see archives 
    and changelog for more)</li>
  <li>Steve Lhomme:&nbsp; ACM wrapper (MS-Windows codec)</li>
  <li><font color="#3366ff">Steve Lhomme:&nbsp; less memory copying on stereo 
    (interleaved) input</font></li>
  <li> <font color="#ff0000">Takehiro Tominaga: Inter-channel masking, enables 
    with --interch x option</font></li>
  <li> For buggy versions of gcc compiler (2.96*), back off on some of the advanced 
    compiler options<br>
  </li>
</ul>
<br>
<h3>LAME 3.91 &nbsp; &nbsp;December 29 2001</h3>
<ul>
  <li><font color="#ff0000">Darin Morrison:&nbsp; Bugfix for --alt-preset (for 
    content with low volume, clean vocals), only important for the "fast standard" 
    preset</font>
  </li>
  <li>Alexander Leidinger:
    <ul>
      <li>add some missing files to the distribution</li>
      <li>add --alt-preset to the man page</li>
    </ul>
  </li>
</ul>
<br>
<h3>LAME 3.90 &nbsp; &nbsp;December 21 2001</h3>
<ul>
  <li><font color="#ff0000">Many small improvements and bug fixes not added to 
    history</font></li>
  <li><font color="#ff0000">John Dahlstrom:&nbsp; more fine tuning on the auto 
    adjustment of the ATH</font></li>
  <li><font color="#3366ff">Robert Hegemann:&nbsp; small speed and quality improvements 
    for the old VBR code (--vbr-old).</font> </li>
  <li><font color="#ff0000">Robert Hegemann:&nbsp; some short block bug fixes</font> 
  </li>
  <li><font color="#ff0000">Robert Hegemann:&nbsp; Big improvements to --vbr-mtrh, 
    now encodes much more frequencies over 16khz</font> </li>
  <li><font color="#ff0000">Robert Hegemann:&nbsp; --vbr-new code disabled (outdated 
    and lower quality) and replaced with --vbr-mtrh (Both --vbr-new and --vbr-mtrh 
    now default to mtrh)</font> </li>
  <li>Robert Hegemann:&nbsp; reordering of --longhelp to give more information, 
    --extrahelp dropped </li>
  <li>Darin Morrison:&nbsp; Totally revamped and extremely high quality unified 
    preset system and other general quality improvements now available with --alt-presets: 
    <ul>
      <li> <font color="#ff0000">some improvements to psychoacoustics (vast improvements 
        over default L.A.M.E. modes) when --alt-preset is used including:</font></li>
      <ul>
        <li> <font color="#ff0000">Improved tuning of short block usage.</font></li>
        <li> <font color="#ff0000">Improved quantization selection usage (the 
          -X modes), now adapts between appropriate modes on the fly. Also helps 
          on "dropout" problems and with pre-echo cases.</font></li>
        <li> <font color="#ff0000">Improved joint stereo usage. Thresholds are 
          better tuned now and fix some "dropout" problems L.A.M.E. suffers from 
          on clips like serioustrouble.</font></li>
        <li> <font color="#ff0000">Improved noise shaping usage. Now switches 
          between noise shaping modes on the fly (toggles -Z on and off when appropriate) 
          which allows lower bitrates but without the quality compromise.</font></li>
        <li> <font color="#ff0000">Clips vastly improved over default L.A.M.E. 
          modes (vbr/cbr/abr, including --r3mix): castanets, florida_seq, death2, 
          fatboy, spahm, gbtinc, ravebase, short, florida_seq, hihat, bassdrum, 
          2nd_vent_clip, serioustrouble, bloodline, and others. No degraded clips 
          known.</font></li>
        <li> VBR bitrates are now more "stable" with less fluctuation -- not dipping 
          too low on some music and not increasing too high unnecessarily on other 
          music. "--alt-preset standard" provides bitrates roughly within the 
          range of 180-220kbps, often averaging close to 192kbps.</li>
      </ul>
      <li> --alt-presets replace the --dm-presets and "metal" preset is removed 
        and replaced with generic abr and cbr presets.</li>
      <li> --alt-preset extreme (note the 'e') replaces xtreme to help eliminate 
        some confusion</li>
      <li> --alt-preset vbr modes now have a fast option which offers almost no 
        compromise in speed.</li>
      <li> --alt-preset standard (and "fast standard") are now much lower in bitrate, 
        matching --r3mix with an overall average, though offering higher quality 
        especially on difficult test samples.</li>
      <li> --alt-presets are no longer just "presets" as in a collection of switches, 
        instead they are now quality "modes" because of special code level tunings 
        (those mentioned above).</li>
      <li> Use --alt-preset help for more information.</li>
    </ul>
  </li>
  <li>Roel VdB: more tuning on the --r3mix preset </li>
  <li>Jon Dee, Roel VdB:&nbsp; INFO tag</li>
  <li>Alexander Leidinger, mp3gain@hotmail.com:&nbsp; added --scale-l and --scale-r 
    to scale stereo channels independantly </li>
  <li>Takehiro Tominaga:&nbsp; <font color="#ff0000">new noise shaping mode, offering 
    more "cutting edge" shaping according to masking, enabled via -q0</font> </li>
  <li>Mark Taylor:&nbsp; More work on --nogap </li>
  <li>Gabriel Bouvigne:&nbsp; Small changes to abr code for more accurate final 
    bitrate </li>
  <li>Gabriel Bouvigne, mp3gain@hotmail.com:&nbsp; Preliminary <a
 href="http://www.replaygain.org"> ReplayGain</a> analysis code added (not functional 
    yet) </li>
  <li>Gabriel Bouvigne, Alexander Leidinger:&nbsp; Documentation updates </li>
  <li>John Dahlstrom, DSPguru@math.com:&nbsp; floating point interface function 
    in the Windows DLL</li>
</ul>
<br>
<h3>LAME 3.89beta&nbsp; &nbsp;July 5 2001</h3>
<ul>
  <li> John Stewart:&nbsp; long filename support for Win9x/NT.</li>
  <li> Takehiro Tominaga:&nbsp; LAME can calculate the CRC of VBR header, so now 
    "lame -pv" works fine.</li>
  <li><font color="#ff0000">Robert Hegemann:&nbsp; Improvements of the new VBR 
    code (--vbr-mtrh).</font></li>
  <li><font color="#3366ff">Robert Hegemann: New VBR code (--vbr-mtrh) is now 
    defaulted to get more feedback. The VBR speed is now on par with CBR. We will 
    use the old VBR code in the release.</font></li>
  <li><font color="#ff0000">Gabriel Bouvigne: Change of the maximum frame size 
    limit. LAME should now be more friendly with hardware players.</font></li>
  <li>Gabriel Bouvigne: Size of VBR is now more balanced according to the -V value.</li>
  <li>Alexander Leidinger: Finished the implementation of the set/get functions.</li>
  <li>John Dahlstrom: LAME now handles 24bits input</li>
  <li>Mark Taylor: bugs in lame --decode causing truncation of mp3 file fixed</li>
  <li>Mark Taylor: preliminary --nogap support</li>
  <li>"Final" API completed: shared library safe! &nbsp;This API is frozen and 
    should be backwords compatiable with future versions of libmp3lame.so, but 
    we will continue to add new functionality. &nbsp;<br>
  </li>
</ul>
<h3> LAME 3.88beta&nbsp;&nbsp; March 25 2001</h3>
<ul>
  <li> <font color="#ff0000">A lot of work that was never added to the History!</font></li>
  <li> <font color="#ff0000">Frank Klemm and Gabriel Bouvigne:&nbsp; New ATH formula.&nbsp; 
    Big improvement for high bitrate encodings.</font></li>
  <li> <font color="#ff0000">Takehiro Tominaga: Temporal masking</font></li>
  <li> <font color="#ff0000">Gabriel Bouvigne/Mark Taylor: auto adjustment of 
    ATH</font></li>
  <li> <font color="#ff0000">Robert Hegemann:&nbsp; Better outer_loop stopping 
    criterion.&nbsp; Enabled with -q2 or better.</font></li>
  <li> <font color="#ff0000">Robert Hegemann/Naoki Shibata:&nbsp; slow/carefull 
    noise shaping.&nbsp; -q3..9:&nbsp; amplify all distorted bands.&nbsp; -q2: 
    amplify distorted bands within 50%.&nbsp; -q1-0:&nbsp; amplify only most distorted 
    band at each iteration.</font></li>
  <li> <font color="#ff0000">Takehiro Tominaga: Interframe, shortblock temporal 
    masking.</font></li>
  <li> Takehiro Tominaga:&nbsp; LAME restructured into a shared library and front 
    end application.&nbsp; Slight changes to the API. More changes are coming 
    to turn LAME into a true shared library (right now you have to recompile if 
    you upgrade the library :-(</li>
  <li> <font color="#000000">Naoki Shibata:</font></li>
  <ul>
    <li> <font color="#ff0000">improvements to psychoacoustics</font><font
 color="#000000"> (--nspsytune)</font></li>
    <li> <font color="#ff0000">BUG in long block pre echo control fixed </font><font color="#000000"> 
      (some out of range array access in M/S psychoacoustics)</font></li>
  </ul>
  <li> <font color="#000000">Ralf Kempkens:&nbsp;&nbsp;&nbsp; Visual Basic Script 
    for lame, suggested to put it on your Windows Desktop and you can drag'n'drop 
    Waves to encode on it.</font></li>
  <li> <font color="#000000">Alexander Stumpf:&nbsp;&nbsp;&nbsp; improved lame.bat 
    for 4Dos users</font></li>
  <li> <font color="#000000">Mark Taylor: Several bugs fixed in the resampling 
    code.</font></li>
  <li> <font color="#000000">Frank Klemm, Robert Hegemann:&nbsp;&nbsp;&nbsp; added 
    assembler code for CPU feature detection on runtime (MMX, 3DNow, SIMD)</font></li>
  <li> <font color="#3366ff">Takehiro Tominaga: 3DNow FFT code.</font></li>
  <li> <font color="#000000">Florian Bome, Alexander Leidinger:&nbsp;&nbsp;&nbsp; 
    more work on configure stuff</font></li>
  <li> <font color="#000000">Alexander Leidinger:&nbsp;&nbsp; automake/libtool 
    generated Makefiles and TONS of other work.</font></li>
  <li> <font color="#000000">Alexander Leidinger:&nbsp;&nbsp; Much work towards 
    shared library style API.</font></li>
  <li> <font color="#000000">Anonymous: New more efficient RTP code.</font></li>
  <li> <font color="#ff0000">Mark Taylor: psycho-acoustic data now computed for 
    all scalefactor bands (up to 24 kHz)</font></li>
  <li> <font color="#ff0000">Mark Taylor, Takehiro Tominaga: All ISO table data 
    replaced by formulas - should improve MPEG2.5 results for which we never had 
    correct table data.</font></li>
</ul>
<h3> LAME 3.87alpha&nbsp; September 25 2000</h3>
<ul>
  <li> Mark Taylor:&nbsp; Bug fixed in LAME/mpglib error recovery when encountering 
    a corrupt&nbsp; MP3 frame during *decoding*.</li>
  <li> Albert Faber:&nbsp; added LayerI+II decoding support</li>
  <li> <font color="#000000">Frank Klemm:&nbsp; added improved CRC calculation</font></li>
  <li> <font color="#000000">Frank Klemm: substantial code cleanup/improvements</font></li>
  <li> Robert Hegemann:&nbsp; Bug fixes</li>
  <ul>
    <li> <font color="#ff0000">in huffman_init</font>, could lead to segmentation 
      faults (only in rare cases, most likely at lower sample rates)</li>
    <li> <font color="#ff0000">M/S switching at lower sample rates</font> (the 
      fact there is no 2nd granule was ignored)</li>
  </ul>
  <li> <font color="#3366ff">Robert Hegemann:&nbsp; speed up in&nbsp; VBR</font></li>
  <li> Jarmo Laakkonen:&nbsp; Amiga/GCC settings for Makefile.unix.</li>
  <li> Magnus Holmgren:&nbsp; README and Makefile for (free) Borland C++&nbsp; 
    compiler. Will also compile lame_enc.dll, but this is untested.</li>
  <li> Florian Bome:&nbsp;&nbsp; LAME finally has a&nbsp; ./configure script!!</li>
</ul>
<h3> LAME 3.86beta&nbsp; August 6 2000</h3>
<ul>
  <li> Christopher Wise:&nbsp; A makefile for DJGPP, the DOS version of gcc.&nbsp; 
    Now most windows users should be able to compile LAME with minimal effort.</li>
  <li> <font color="#ff0000">Robert Hegemann:&nbsp; old VBR:&nbsp;&nbsp; fixed 
    some bugs and Takehiro's scalefac_scale feature (not yet on by&nbsp; default.)&nbsp; 
    older LAME versions did not allow to spent more than 2500 bits of 4095 possible 
    bits to a granule per channel, now fixed.</font></li>
  <li> Robert Hegemann:&nbsp; new VBR:&nbsp;&nbsp; analog silence treatment like 
    in old VBR</li>
  <li> William Welch:&nbsp; Improved options for Linux/Alpha gcc and ccc compilers 
    in Makefile.</li>
  <li> Mathew Hendry:&nbsp; setting appropriate CRC bit for additional Xing-VBR 
    tagging frame</li>
  <li> Don Melton:&nbsp; added ID3 version 2 TAG support</li>
  <li> <font color="#000000">John Dahlstrom: fixed bug allowing timing information 
    (for status in command line encoder) to overflow.</font></li>
  <li> <font color="#000000">Tamito KAJIYAMA, Fixed several bugs in the LAME/Vorbis 
    interface.</font></li>
  <li> <font color="#000000">Mark Taylor:&nbsp; lame --decode will recognize <a href="http://albumid.cjb.net">Album 
    ID tags</a></font></li>
  <li> <font color="#ff0000">Naoki Shibata:&nbsp; Additive masking and other improvements 
    to psycho acoustics.&nbsp; (not yet on by default)</font></li>
</ul>
<h3> LAME 3.85beta&nbsp;&nbsp; July 3 2000</h3>
<ul>
  <li> <font color="#ff0000">Takehiro Tominaga:&nbsp; mid/side stereo demasking 
    thresholds updated.</font></li>
  <li> Takehiro Tominaga: New short block MDCT coefficient data structure.&nbsp; 
    Should allow for future speed improvements.</li>
  <li> Robert Hegemann:&nbsp; fixed bug in old VBR routine, the --noath mode messed 
    up the VBR routine resulting in very large files</li>
  <li> Robert Hegemann: found bugs in some sections when using 32 bit floating 
    point.&nbsp; Default is now back to 64bit floating point.</li>
  <li> <font color="#ff0000">Takehiro Tominaga:&nbsp; Modified PE formula to use 
    ATH.</font></li>
  <li> <font color="#000000">S.T.L.:&nbsp; README.DJGPP - instructions for compiling 
    LAME with DJGPP, the dos version of gcc.</font></li>
</ul>
<h3> LAME 3.84beta&nbsp; June 30&nbsp; 2000</h3>
<ul>
  <li> Mark Weinstein:&nbsp; .wav file output (with --decode option) was writing 
    the wrong filesize in the .wav file.&nbsp; Now fixed.</li>
  <li> Mark Taylor:&nbsp; (optional) Vorbis support, both encoding and decoding.&nbsp; 
    LAME can now produce .ogg files, or even re-encode your entire .ogg collection 
    into mp3.&nbsp;&nbsp; (Just kidding: it is always a bad idea to convert from 
    one lossy format to another)</li>
  <li> ?: Bug fixed causing VBR to crash under windows.&nbsp;&nbsp; (pretab[] 
    array overflow)</li>
  <li> Sergey Sapelin: Another bug found in the mpg123 MPEG2 tables.&nbsp; Now 
    fixed for the mpg123 based decoder in LAME.</li>
  <li> Marco Remondini:&nbsp; VBR histogram works in win32.&nbsp; compile with 
    -DBRHIST -DNOTERMCAP</li>
  <li> <font color="#ff0000">Takehiro Tominaga:&nbsp; LAME CBR will now use scalefac_scale 
    to expand the dynamic range of the scalefactors.</font></li>
  <li> <font color="#000000">Iwasa Kazmi:&nbsp; Library improvements:&nbsp; exit()'s, 
    printf, fprintf's are being replaced by interceptable macros.</font></li>
</ul>
<h3> LAME 3.83beta&nbsp; May 19&nbsp; 2000</h3>
<ul>
  <li> <font color="#ff0000">Mark Taylor:&nbsp; Bug in buffering routines:&nbsp; 
    in some cases, could cause MDCT&nbsp; to read past end of buffer.&nbsp; Rare 
    in MPEG2, even more rare for MPEG1, but potentially serious!</font></li>
  <li> Mark Taylor:&nbsp; MDCT/polyphase filterbank was not being "primed" properly.&nbsp; 
    Does not effect output unless you set the encoder delay lower than the default 
    of 576 samples.</li>
  <li> <font color="#ff0000">Mark Taylor:&nbsp; "vdbj" and "Caster"&nbsp; found 
    several VBR bugs (now fixed):&nbsp;&nbsp; 1.&nbsp; Analog silence detection 
    only checked frequencies up to 16 kHz.&nbsp; 2.&nbsp; VBR mode could still 
    somehow avoid -F mode.&nbsp; 3.&nbsp; VBR mode would ignore noise above 16 
    kHz (scalefactor band 22), Now calc_noise1 will compute the noise in this 
    band when in VBR mode.&nbsp; Not calculated in CBR&nbsp; mode since CBR algorithm 
    has no way of using this information.</font></li>
  <li> Mark Taylor:&nbsp; scalefactor band 22 info (masking(=ATH),&nbsp; noise 
    and energy) now displayed in frame analyzer.</li>
  <li> <font color="#ff0000">VBR code ATH tuning was disabled by accident in 3.81, 
    now fixed.</font></li>
  <li> <font color="#000000">Mark Taylor:&nbsp; lame --decode will produce .wav 
    files.&nbsp; (oops - size is off by a factor of 4)</font></li>
</ul>
<h3> LAME 3.82beta&nbsp;&nbsp; May 11 2000</h3>
<ul>
  <li> Robert Hegemann:&nbsp; Fixed bug in high bitrate joint stereo encodings.</li>
  <li> <font color="#3366ff">Naoki Shibata:&nbsp; new long block MDCT routine</font></li>
</ul>
<h3> LAME 3.81beta&nbsp; May 8 2000</h3>
<ul>
  <li> all ISO code removed!</li>
  <li> <font color="#3366ff">Takehiro Tominaga and Naoki Shibata:&nbsp; new window 
    subband routines.</font></li>
  <li> <font color="#000000">Naoki Shibata:&nbsp; Bug fix in mpglib (decoding) 
    lib:&nbsp; in some cases, MDCT coefficients from previous granule was incorrectly 
    used for the next granule.</font></li>
  <li> <font color="#ff0000">ISO 7680 bit buffer limitation removed.&nbsp; It 
    can be reactivated with "--strictly-enforce-ISO"&nbsp; Please report any trouble 
    with high bitrates.</font></li>
</ul>
<h3> LAME 3.80beta&nbsp; May 6 2000</h3>
<ul>
  <li> <font color="#ff0000">Takehiro Tominaga:&nbsp; more efficient and faster 
    huffman encoding!</font></li>
  <li> <font color="#ff0000">Takehiro Tominaga and Mark Taylor:&nbsp; much improved 
    short block compression!</font></li>
  <li> <font color="#000000">Tomasz Motylewski and Mark Taylor:&nbsp; MPEG2.5 
    now supported!</font></li>
  <li> <font color="#000000">Mark Taylor: incorporated Takehiro's bitstream.c!&nbsp; 
    bitstream.c used by default, but old ISO bitstream code can also be used.</font></li>
  <li> <font color="#ff0000">Scott&nbsp; Manley and Mark Taylor:&nbsp; good resampling 
    routine finaly in LAME.&nbsp; uses a 19 point FIR filter with Blackman window.&nbsp; 
    Very slow for non integer resampling ratios.</font></li>
  <li> <font color="#000000">Iwasa Kazmi:&nbsp; fixed SIGBUS error:&nbsp; VBR 
    and id3 tags were using data after it was free()'d.</font></li>
  <li> <font color="#ff0000">Robert Hegemann:&nbsp; Improved VBR tuning.&nbsp; 
    #define RH_QUALITY_CONTROL and #RH_SIDE_VBR now the defaults.</font></li>
  <li> <font color="#000000">Robert Hegemann:&nbsp;&nbsp; LAME version string 
    now added to ancillary data.</font></li>
  <li> Kimmo Mustonen:&nbsp; VBR histogram support for Amiga.</li>
  <li> Casper Gripenberg:&nbsp; VBR stats (but not histogram) for DOS verson.</li>
  <li> Robert Hegemann:&nbsp; rare VBR overflow bug fixed.</li>
  <li> Zack:&nbsp; -F option strictly enforces the VBR min bitrate.&nbsp; Without 
    -F, LAME will ignore the minimum bitrate when encoding analog silence.</li>
  <li> Shawn Riley:&nbsp; User can now specify a compression ratio (--comp &lt;arg&gt;) 
    instead of a bit rate.&nbsp; Default settings based on a compression ratio 
    of 11.0</li>
  <li> Mark Taylor:&nbsp; free format bitstreams can be created with --freeformat, 
    and specify any integer bitrate from 8 to 320kbs with -b.</li>
  <li> Mark Taylor: lame be used as a decoder (output raw pcm only):&nbsp; lame 
    --decode input.mp3 output.pcm</li>
</ul>
<h3> LAME 3.70&nbsp;&nbsp; April 6 2000</h3>
<ul>
  <li> "LAME 3.69beta" becomes LAME 3.70 "stable"</li>
</ul>
<h3> LAME 3.69beta&nbsp;&nbsp; April 6 2000</h3>
<ul>
  <li> "spahm":&nbsp; default mode selection bug fixed.&nbsp; In some cases, lame 
    was defaulting to regular stereo instead of jstereo when the user did not 
    specify a mode.</li>
</ul>
<h3> LAME 3.68beta&nbsp; April 4 2000</h3>
<ul>
  <li> Mark Taylor: mono encoding bug in DLL fixed.</li>
  <li> Ingo Saitz: bug in --cwlimit argument parsing fixed.</li>
  <li> <font color="#ff0000">Scott Manly: bug in 4-point resample code fixed.</font></li>
</ul>
<h3> LAME 3.67beta&nbsp; March 27 2000</h3>
<ul>
  <li> <font color="#ff0000">Robert Hegemann:&nbsp; jstereo now enabled for MPEG2 
    encodings</font></li>
  <li> Mark Taylor: old M/S stereo mode which used L/R maskings has been removed.</li>
  <li> Mark Taylor: Xing MPEG2 VBR headers now working.</li>
  <li> <font color="#ff0000">Mark Taylor:&nbsp; When quantized coefficients are 
    all 0 in a band, set scalefactors to 0 also to save a few bits.</font></li>
  <li> <font color="#000000">Ingo Saitz:&nbsp; Problems with framesize calculation 
    when using -f fast-math option fixed.</font></li>
</ul>
<h3> LAME 3.66beta March 21 2000</h3>
<ul>
  <li> Bug fixes in BladeEnc DLL, possible click in last mp3 frame, VBR historgram 
    display, byteswapping option, ASM quantize routines work for both float and 
    double.</li>
</ul>
<h3> LAME 3.65beta&nbsp;&nbsp; March 17 2000</h3>
<ul>
  <li> Enabled ASM version of quantize_xrpow() - accidently disabled in lame3.64.</li>
</ul>
<h3> LAME 3.64beta&nbsp; March 16 2000</h3>
<ul>
  <li> Don Melton:&nbsp; id3v1.1 tags &amp; id3 bugfixes</li>
  <li> <font color="#ff0000">Gabriel Bouvigne:&nbsp; L/R matching block type fix</font></li>
  <li> <font color="#ff0000">Bug fixed which was allowing quantized values to 
    exceed the maximum when not using -h</font></li>
  <li> <font color="#3366ff">Mark Taylor: Fitlers based on polyphase filterbank.&nbsp; 
    should be slightly better since the responce is independent of the blocktype, 
    and they are slightly faster.</font></li>
  <li> Mark Taylor: API:&nbsp; the API changed slightly - and this should be the 
    final version. There is a new routine: lame_encode_buffer() which takes an 
    arbritray sized input buffer, resamples &amp; filters if necessary, encodes, 
    and returns the mp3buffer.&nbsp; There are also several new #defines, so it 
    is possible to compile a simple encoding library with no decoding or file 
    I/O or command line parsing.&nbsp; see the file API for details.</li>
  <li> Mark Taylor: MSVC stuff:&nbsp; lame.exe (with and without the frame analyzer) 
    and the CDex lame_enc.dll</li>
  <br>
  should compile under MSVC.&nbsp; The MSVC5 project files may need some tweaking.&nbsp; 
  In particular, <br>
  you need to make sure LAMEPARSE, LAMESNDFILE and HAVEMPGLIB <br>
  are defined.&nbsp; (and HAVEGTK for the GTK stuff). 
</ul>
<h3> LAME 3.63beta&nbsp; February 20&nbsp; 2000</h3>
<ul>
  <li> Robert Hegemann:&nbsp; FPE with -h fixed?</li>
  <li> Mathey Hendry:&nbsp; FPE error catching for Cygwin, FPE fix for vbr mode 
    and output to /dev/null</li>
  <li> Jeremy Hall:&nbsp; Fixed problems with input files where the number of 
    samples is not known.</li>
  <li> <font color="#3366ff">Mathew Hendry:&nbsp; ASM quantize_xrpow() for GNU 
    i386</font></li>
  <li> <font color="#3366ff">Wilfried Behne&nbsp; quantize_xrpow ()for PowerPC 
    and non-ASM</font></li>
  <li> <font color="#3366ff">Takehiro Tominaga:&nbsp; GOGO FFTs&nbsp; (not yet 
    used?)</font></li>
  <br>
  &nbsp; 
</ul>
<h3> LAME 3.62beta&nbsp;&nbsp; February 9 2000</h3>
<ul>
  <li> <font color="#000000">Iwasa Kazmi:&nbsp; frame analyzer short block display 
    of single subblocks (press 1,2&nbsp; or 3)</font></li>
  <li> <font color="#000000">Ingo Saitz:&nbsp; --help option added, with output 
    to stdout</font></li>
  <li> <font color="#ff0000">Alfred Weyers: short block AAC spreading function 
    bug fixed</font></li>
  <li> <font color="#3366ff">Takehiro Tominaga:&nbsp; new scalefac data structure 
    - improves performance!</font></li>
  <li> <font color="#ff0000">Lionel Bonnet:&nbsp; Bug fixed in MPEG2 scalefactor 
    routine: scalefactors were being severly limited.</font></li>
  <li> <font color="#3366ff">Takehiro Tominaga:&nbsp; faster FFT routines from.&nbsp; 
    These routines are also compatible with the GOGO routines, in case someone 
    is interested in porting them back to LAME.</font></li>
  <li> <font color="#3366ff">Sigbj&oslash;rn Skj&aelig;ret, Takehiro Tominaga:&nbsp; 
    faster pow() code.</font></li>
  <li> <font color="#ff0000">Joachim Kuebart:&nbsp; Found some unitialized variables 
    that were effecting quality for encodings which did not use the -h option 
    (now fixed).</font></li>
  <li> Mark Taylor: More modularization work.&nbsp;&nbsp; It is now possible to 
    use LAME as a library where you can set the encoding parameters directly and 
    do your own file i/o.&nbsp;&nbsp; The calling program is now it's own mp3 
    output.&nbsp; For an example of the LAME API, see main.c, or mp3rtp.c or mp3x.c.&nbsp; 
    These can all be compiled as stand alone programs which link with libmp3lame.a.</li>
  <li> Felix vos Leitner:&nbsp; mp3rtp fixes.&nbsp; mp3rtp is a standalone program 
    which will encode and stream with RTP.</li>
  <li> Robert Hegemann:&nbsp; Information written to stderr displaying exactly 
    which type of lowpass filter (if any) is being used.</li>
  <li> Iwasa Kazmi:&nbsp; mpglib (the mpg123 decoder) scsfi decoding fixes.</li>
  <li> Takehiro Tominaga:&nbsp; More mpglib scsfi decoding fixes.</li>
</ul>
<h3> LAME 3.61beta&nbsp; January 14 2000</h3>
<ul>
  <li> <font color="#ff0000">Mark Taylor: Fixed bug with lowpass filters when 
    using VBR with a 64kbs or lower min bitrate setting.</font></li>
  <li> <font color="#ff0000">Takehiro Tominaga:&nbsp; more efficient huffman encoding 
    splitting.</font></li>
</ul>
<h3> LAME 3.60beta&nbsp;&nbsp; January 9 2000</h3>
<ul>
  <li> Mark Taylor: Distribution now comes with self test.&nbsp; Needs work to 
    be automated, see 'make test' in Makefile.</li>
  <li> <font color="#ff0000">Mark Taylor: AAC spreading function now the default</font></li>
  <li> Gabriel Bouvigne: updated HTML docs</li>
  <li> Felix von Leitner: compute correct file length from Xing header (if present) 
    when input file is a mp3 file</li>
  <li> Felix von Leitner: mp3rtp (standalone) program now included.&nbsp; Not 
    yet tested.&nbsp; mp3rtp ip:port:ttl&nbsp; &lt;infile&gt;&nbsp;&nbsp; /dev/null 
    will stream directly to ip:port using&nbsp; RTP.</li>
  <br>
  &nbsp; 
</ul>
<h3> LAME 3.59beta&nbsp; January 4 2000</h3>
<ul>
  <li> Takehiro Tominaga:&nbsp; --noath option.&nbsp; Disables ATH maskings.</li>
  <li> Gabriel Bouvigne:&nbsp; updated HTML docs.</li>
  <li> Iwasa Kazmi:&nbsp; makefile fixes</li>
  <li> Mark Taylor:&nbsp; Fixed bug where first frame of data was always overwritten 
    with 0's.&nbsp; Thanks to 'gol'</li>
  <li> <font color="#ff0000">Mark Taylor:&nbsp; bug fixes in mid/side masking 
    ratios (thanks to Menno Bakker)</font></li>
  <li> Mark Taylor:&nbsp; replaced norm_l, norm_s table data with formulas.</li>
</ul>
<h3> LAME 3.58beta&nbsp; December 13 1999</h3>
<ul>
  <li> <font color="#ff0000">Segher Boessenkool:&nbsp; More accurate quantization 
    procedure!&nbsp; Enabled with -h.</font></li>
  <li> <font color="#3366ff">Mathew Hendry, Acy Stapp and Takehiro Tominaga: ASM 
    optimizations for quantize_xrpow and quantize_xrpow_ISO.</font></li>
  <li> Chuck Zenkus:&nbsp; "encoder inside" logo on web page</li>
  <li> Mark Taylor:&nbsp;&nbsp; a couple people have asked for this.&nbsp;&nbsp; 
    Allow LAME to overide VBR_min_bitrate if analog_silence detected.&nbsp;&nbsp; 
    Analog_silence defined a la Robert:&nbsp;&nbsp; energy&nbsp; &lt; ATH.</li>
  <li> An Van Lam: Valid bitrates were being printed for layer 2, not layer 3!</li>
  <li> Ethan Yeo:&nbsp; Makefile.MSVC updated</li>
  <li> Mark Stephens:&nbsp; updated all MSVC project files</li>
  <li> Robert Hegemann:&nbsp; lowpass and highpass filters can be enabled with 
    --lowpass, --highpass</li>
  <li> <font color="#ff0000">Mark Taylor:&nbsp; MS switching is now smoother: 
    ms_ratio average over 4 granules</font></li>
  <li> <font color="#ff0000">Takehiro Tominaga:&nbsp; Scalefactor pre-emphasis 
    fixed (and now turned back on)</font></li>
  <li> <font color="#ff0000">Takehiro Tominaga:&nbsp; Bug in M/S maskings:&nbsp; 
    switch to turn on stereo demasking code was buggy.</font></li>
</ul>
<h3> LAME 3.57beta&nbsp; November 22 1999</h3>
<ul>
  <li> Sigbj&oslash;rn Skj&aelig;ret, patch to allow encoding from 8bit input 
    files when using LIBSNDFILE</li>
  <li> Mark Taylor: Automatic downsampling to nearest valid samplerate.</li>
  <li> Mark Taylor: Scalefactor bands demarked on MDCT plot in frameanalyzer</li>
  <li> Mark Taylor: Scalefactor preemphasis disabled for now.&nbsp;&nbsp; The 
    algorithm was often doing more harm than good.</li>
</ul>
<h3> LAME 3.56beta&nbsp; November 19 1999</h3>
<ul>
  <li> Kimmo Mustonen: portabilty code cleanup.</li>
  <li> Vladimir Marek: id3 genre patch.</li>
  <li> Conrad Sanderson: new applypatch script.</li>
  <li> Mark Taylor: Initial window type now "STOP_TYPE" to reduce initial attenuation.&nbsp; 
    This is needed because the new encoder delay is so short.&nbsp; With a NORM_TYPE, 
    the first 240 samples would be attenuated.</li>
  <li> Mark Taylor: Padding at end of file now adjusted (hopefully!) to produce 
    as little padding as possible while still guarantee all input samples are 
    encoded.</li>
  <li> <font color="#ff0000">Takehiro Tominaga:&nbsp; Reduced shortblock extra 
    bit allocation formulas by 10% since new huffman coding is at least 10% more 
    efficient.</font></li>
</ul>
<h3> LAME 3.55beta&nbsp; November 11 1999</h3>
<ul>
  <li> Albert Faber:&nbsp; updated BladeEnc.dll</li>
  <li> Mark Taylor: Simple lowpass filter added to linear downsampling routine.</li>
  <li> Nils Faerber: updated man page.</li>
  <li> Mark Taylor: All floating point variables are delcared FLOAT or&nbsp; FLOAT8.&nbsp; 
    Change the definition of FLOAT8 in machine.h to run at 32bit preceision.</li>
  <li> Mark Taylor: Bug (introduced in 3.54beta) in stereo-&gt;mono downsampling 
    fixed.</li>
</ul>
<h3> LAME 3.54beta&nbsp; November 8 1999</h3>
<ul>
  <li> Mark Taylor: Encoder delay is now 48 samples.&nbsp; Can be adjusted to 
    1160 to sync with FhG (see ENCDELAY in encoder.h)&nbsp; This is kind of amazing, 
    since if Takehiro put his MDCT/filterbank routine in a decoder, we could have 
    a total&nbsp; delay of only 96 samples.</li>
  <li> <font color="#ff0000">Mark Taylor: More inconstancies found and fixed in 
    MPEG2 tables.</font></li>
  <li> Mark Taylor: Resampling from an MP3 input file now works.&nbsp; But we 
    still dont have a lowpass filter so dont expect good results.</li>
</ul>
<h3> LAME 3.53beta&nbsp; November 8 1999</h3>
<ul>
  <li> <font color="#3366ff">Takehiro Tominaga:&nbsp; Fixed MPEG2 problem in new 
    MDCT routines.&nbsp; Takehiro's combined filterbank/MDCT routine is now the 
    default.&nbsp; Removes all buffering from psymodel.c and the filterbanks/MDCT 
    routines.</font></li>
</ul>
<h3> LAME 3.52beta&nbsp; November 8 1999</h3>
<ul>
  <li> By permission of copyright holders of all GPL code in LAME,&nbsp; all GPL 
    code is now released under a modified version of the LGPL (see the README 
    file)</li>
  <li> By popular demand, all C++ comments changed to C style comments</li>
  <li> Mark Taylor: Linear resampling now works.&nbsp; Use --resample to set an 
    output samplerate different from the input samplerate.&nbsp; (doesn't seem 
    to work with mp3 input files, and there is no lowpass filter, so dont expect 
    good results just yet)</li>
  <li> <font color="#3366ff">Takehiro Tominaga:&nbsp; Faster Huffman encoding 
    routines</font></li>
</ul>
<font color="#3366ff">The following changes are disabled because of MPEG2 problems.&nbsp; 
But to try them, set MDCTDELAY=48 in encoder.h, instead of MDCTDELAY=528.:</font> 
<ul>
  <li> <font color="#3366ff">Takehiro Tominaga:&nbsp; New MDCT routines with shorter 
    delay (48 samples instead of 528) and even faster than the old routines.</font></li>
  <li> <font color="#3366ff">Takehiro Tominaga:&nbsp; Removed extra buffering 
    in psymodel.c</font></li>
</ul>
<h3> LAME 3.51&nbsp; November 7 1999</h3>
<ul>
  <li> Takehiro Tominaga: Bug in quantize.c absolute threshold of hearing calculation 
    for non-44.1 kHz input files.</li>
</ul>
<h3> LAME 3.50&nbsp; November 1 1999</h3>
<ul>
  <li> LAME 3.37beta becomes official LAME 3.50 release</li>
</ul>
<h3> LAME 3.37beta&nbsp; November 1 1999</h3>
<ul>
  <li> <font color="#ff0000">Lionel Bonnet:&nbsp; Found severe bug in MPEG2 Short 
    block SNR.</font></li>
  <li> Sergey Sapelin:&nbsp; VBR Toc improvement.</li>
  <li> Sergey Dubov: fskip() routine</li>
  <li> Conrad Sanderson: replacement for filterbank.c.&nbsp;&nbsp; Not much faster 
    but amazingly simpler.</li>
</ul>
<h3> LAME 3.36beta&nbsp; October 25 1999</h3>
<ul>
  <li> Albert Faber:&nbsp; more MSVC and BladeDLL updates</li>
  <li> Kimmo Mustonen:&nbsp; Much code cleanup and Amiga updates</li>
  <li> Anton Oleynikov: Borland C updates</li>
  <li> Mark Taylor: More stdin fixes:&nbsp; For some reason, forward fseek()'s 
    would fail when used on pipes even though it is okay with redirection from 
    "&lt;". So I changed all the forward fseek()'s to use fread().&nbsp; This 
    should improve stdin support for wav/aiff files.&nbsp; If you know the input 
    file is raw pcm, you can still use the '-r' option to avoid *all* seeking 
    of any kind.</li>
  <br>
  &nbsp; 
</ul>
<h3> LAME 3.35beta&nbsp; October 21 1999</h3>
<ul>
  <li> <font color="#ff0000">Leonid Kulakov:&nbsp; Serious bug in MPEG2 scalefactor 
    band tables fixed.</font></li>
  <li> Portability patches from:&nbsp; Anton Oleynikov, Sigbj&oslash;rn Skj&aelig;ret, 
    Mathew Hendry, Richard Gorton</li>
  <li> Alfred Weyers: compiler options, updated timestatus.</li>
  <li> Albert Faber:&nbsp; BladeDll and other updates (new machine.h).</li>
  <li> Monty:&nbsp; updated Makefile to fix gcc inline math bug.</li>
  <br>
  &nbsp; 
</ul>
<h3> LAME 3.34beta&nbsp; October 12 1999</h3>
<ul>
  <li> <font color="#ff0000">Mark Taylor: Bug fixed:&nbsp; minimum bitrate in 
    VBR mode could be ignored for a few frames.</font></li>
  <li> <font color="#ff0000">Mark Taylor: New (minor) VBR tunings.</font></li>
  <li> Tim Ruddick: New wav/aiff header parsing routines.&nbsp; Better parsing 
    and fewer fseek()'s.</li>
  <li> Anton Oleynikov:&nbsp; patches to work with Borland C</li>
  <li> <font color="#ff0000">Gabriel Bouvigne:&nbsp; Experimental voice option 
    enabled with --voice</font></li>
</ul>
<h3> LAME 3.33beta&nbsp; October 11 1999</h3>
<ul>
  <li> <font color="#ff0000">Robert Hegemann: RH VBR mode now the default and 
    only VBR mode.&nbsp; The new code will always quantize to 0 distortion and 
    the quality is increased by reducing the masking from the psy-model.&nbsp; 
    -X0 is still the default for now.</font></li>
  <li> <font color="#ff0000">Robert Hegemann: new -X5 mode</font></li>
  <li> Mathew Hendry: New timing code, removes the need for HAVETIMES</li>
  <li> <font color="#3366ff">Mathew Hendry:&nbsp; assembler quantize_xrpow for 
    Windows</font></li>
  <li> Iwasa Kazmi:&nbsp; stdin/stdout patch for Windows</li>
  <li> Mark Taylor: New option: "--athonly" will ignore the psy-model output and 
    use only the absolute threshold of hearing for the masking.</li>
  <br>
  &nbsp; 
</ul>
<h3> LAME 3.32beta&nbsp; October 8 1999</h3>
<ul>
  <li> <font color="#3366ff">Takehiro Tominaga:&nbsp; faster long block spreading 
    function convolution for non 44.1 kHz sampling frequencies, and faster short 
    block spreading function convolution for all sampling frequencies.</font></li>
  <li> <font color="#ff0000">Takehiro Tominaga:&nbsp; Completly rewritten huffman 
    table selection and count_bits().&nbsp; More efficient table selection results 
    in many more bits per frame.</font></li>
  <li> <font color="#ff0000">Takehiro Tominaga:&nbsp; More efficient scalefac 
    compress setting.</font></li>
  <li> <font color="#3366ff">Mike Cheng: new calc_noise2()</font></li>
  <li> Alfred Weyers: patch for timestatus() seconds rollover</li>
  <br>
  &nbsp; 
</ul>
<h3> LAME 3.31beta&nbsp; September 28 1999</h3>
<ul>
  <li> Albert Faber:&nbsp; updated his BladeDLL code.&nbsp; This allows LAME to 
    be compiled into a BladeEnc compatiable .dll.</li>
  <li> <font color="#3366ff">Mike Cheng: faster l3psycho_ener() routine.</font></li>
  <li> Sigbj&oslash;rn Skj&aelig;ret: more code cleanup.</li>
</ul>
<h3> LAME&nbsp; 3.30beta&nbsp; September 27 1999</h3>
<ul>
  <li> Conrad Sanderson:&nbsp; ID3 tag code added (type 'lame' for instructions)</li>
  <li> new mdct.c from Mike Cheng (no faster, but much cleaner code)</li>
  <li> Mathew Hendry: Microsoft nmake makefile and a couple other changes for 
    MSVC</li>
  <li> More modulization work:&nbsp; One input sound file interface handles mp3's, 
    uncompressed audio, with or without LIBSNDFILE.&nbsp; Fixes (hopefully) a 
    bunch of file I/O bugs introduced in 3.29 (Mark Taylor)</li>
  <li> LAME will now print valid samplerate/bitrate combinations (Mark Taylor)</li>
  <li> stdin/stdout fix for OS/2 (Paul Hartman)</li>
  <li> For mp3 input files, totalframes estimated based on filesize and first 
    frame bitrate. (Mark Taylor)</li>
  <li> Updated all functions with new style prototypes.&nbsp; (Sigbj&oslash;rn 
    Skj&aelig;ret)</li>
  <br>
  &nbsp; 
</ul>
<h3> LAME 3.29beta&nbsp; September 21 1999</h3>
<ul>
  <li> <font color="#ff0000">Bug in bigv_bitcount fixed.&nbsp; Loop.c was overestimating 
    the number of bits needed, resulting in wasted bits every frame.&nbsp; (Leonid 
    A. Kulakov)</font></li>
  <li> <font color="#ff0000">Bug in *_choose_table() fixed&nbsp;&nbsp; These routines 
    would not sellect the optimal Huffman table in some cases.&nbsp;&nbsp; (Leonid 
    A. Kulakov)</font></li>
  <li> <font color="#ff0000">Tuning of ATH normalization (macik)</font></li>
  <li> Removed unused variables and fixed function prototypes (Sigbj&oslash;rn 
    Skj&aelig;ret)</li>
  <li> Sami Farin sent a&nbsp; .wav&nbsp; file that&nbsp; LAME built in support 
    choked on.&nbsp;&nbsp; I added a slightly more sophisticated wav header parsing 
    to handle this, but if you have trouble, use libsndfile.</li>
  <li> Resampling hooks and options added.&nbsp; Buffering and resampling routines 
    need to be written.</li>
  <li> LAME will now take an mp3 file as input.&nbsp; When resampling code is 
    working, LAME will be able to (for example) convert a high bitrate stereo 
    mp3 to a low bitrate mono mp3 for streaming.</li>
</ul>
<h3> LAME 3.28beta&nbsp; September 15 1999</h3>
<ul>
  <li> <font color="#ff0000">Serious bug fixed in high frequency MDCT coefficients.&nbsp; 
    Huffman coding was reversing the order of the count1 block quadruples.&nbsp;&nbsp; 
    (Leonid A. Kulakov)</font></li>
  <li> nint() problems under Tru64 unix fixed and preprocessor variable HAVE_NINT 
    removed.&nbsp; (Bob Bell)</li>
  <li> Compiler warning fixes and code cleanup&nbsp;&nbsp; (Sigbj&oslash;rn Skj&aelig;ret, 
    Lionel Bonnet)</li>
  <li> USAGE file now includes suggestions for downsampling.&nbsp; For low bitrate 
    encodings, proper downsampling can give dramatically better results.&nbsp; 
    (John Hayward-Warburton)</li>
</ul>
<h3> LAME 3.27beta&nbsp; September 12 1999</h3>
<ul>
  <li> Several bugs in encode.c and l3bitstream.c fixed by Lionel Bonnet.</li>
  <li> Bugs in new VBR (#define RH) formula for mono input file and mid/side encoding 
    fixed.</li>
</ul>
<h3> LAME 3.26beta&nbsp; September 10 1999</h3>
<ul>
  <li> The "-m m" option (mono .mp3 file) will automatically mix left and right 
    channels if the input file is stereo.&nbsp; (Alfred Weyers)</li>
  <li> <font color="#ff0000">New quant_compare algorithm (method for deciding 
    which of two quantizations is better) enabled with -X4 (Greg Maxwell)</font></li>
  <li> <font color="#ff0000">New mid/side VBR bit allocation formula.&nbsp; Mid 
    channel bits are set by the quality requirements, and then the side channel 
    uses a reduced number of bits (in a proportion coming from the fixed bitrate 
    code).&nbsp; This might not be optimal, but it should be pretty good and no 
    one knows what the optimal solution should be.&nbsp; (Greg Maxwell)</font></li>
  <li> <font color="#ff0000">New VBR (#define RH) tunings based on detailed listening 
    tests by Macik and Greg Maxwell.</font></li>
  <li> Sigbj&oslash;rn Skj&aelig;ret fixed several compiler warnings (which turned 
    out to be potential bugs)</li>
  <li> Takehiro Tominaga fixed a low bitrate bug in reduce_side()</li>
  <li> Alfred Weyers fixed some buffer overflows.</li>
  <li> <font color="#ff0000">New ATH (absolute threshold of hearing) formula replaces 
    buggy ISO code, and adds&nbsp; analog silence treatment&nbsp; (removal of 
    coefficients below below ATH).&nbsp;&nbsp; These are turned on by default 
    but have not been fully tested.&nbsp; (Robert Hegemann)</font></li>
  <li> <font color="#ff0000">Bug in short block spreading function fixed.&nbsp; 
    (Robert Hegemann)</font></li>
</ul>
<h3> LAME 3.25beta&nbsp; August 22 1999</h3>
<ul>
  <li> Sigbj&oslash;rn Skj&aelig;ret fixed a zero byte malloc call.&nbsp; This 
    bug&nbsp; was introduced in 3.24 and causes problems on non Linux systems.</li>
  <li> Bit allocation routines would sometimes allocate more than 4095 bits to 
    one channel of one granule.&nbsp; A couple of people reported problems that 
    might be caused by this, especially at higher bitrates.</li>
  <li> Nils Faerber updated the man page and fixed many of the compiler warnings.</li>
  <br>
  &nbsp; 
</ul>
<h3> LAME 3.24beta&nbsp;&nbsp; August 15 1999</h3>
<ul>
  <li> This release contains the following new code (for developers) which is 
    disabled by default:</li>
  <li> Robert Hegemann:&nbsp; Completely overhauled VBR code.&nbsp; Now computes 
    exact number of bits required for the given qualty and then quantized with 
    the appropriate bitrate.</li>
  <li> Several new quantization quality measures.</li>
</ul>
<h3> LAME 3.23beta&nbsp; August 8 1999</h3>
<ul>
  <li> Very nice continuously updated VBR histogram display from Iwasa Kazmi.&nbsp; 
    (disabled with --nohist).</li>
  <li> More modulerization work.&nbsp; The encoding engine can now be compiled 
    into libmp3lame, but the interface is awkward.</li>
  <li> <font color="#ff0000">Bug fixed in FFT Hann window formula (Leonid A. Kulakov).</font></li>
  <li> New LAME logo on the download page.&nbsp; Created by Chris Michalisles.</li>
  <li> <font color="#ff0000">Several VBR algorithm improvements from Robert Hegemann.&nbsp; 
    New quantization noise metrics and VBR quality measure takes into account 
    mid/side encoding.&nbsp; Should produce smaller files with the same quality, 
    especially when using jstereo.</font></li>
</ul>
<h3> LAME 3.22beta&nbsp; July 27 1999</h3>
<ul>
  <li> Downsampling (stereo to mono) bug with MPEG2 fixed.&nbsp; (Mike Oliphant)</li>
  <li> Downsampling now merges L &amp; R channels - before it only took the L 
    channel.</li>
  <li> More modularization and code cleanup from Albert Faber and myself.</li>
  <li> Input filesize limit removed for raw pcm input files.&nbsp; For other file 
    types, LAME will still only read the first 2^32 samples, (27 hours of playing 
    time at 44.1 kHz).</li>
  <br>
  &nbsp; 
</ul>
<h3> LAME 3.21beta&nbsp; July 26 1999</h3>
<ul>
  <li> <font color="#ff0000">Correct Mid/Side masking thresholds for JSTEREO mode!&nbsp; 
    This is enabled with -h.&nbsp; It&nbsp; makes LAME about 20% slower since 
    it computes psycho-acoustics for L,R Mid and Side channels.</font></li>
  <li> <font color="#ff0000">"Analog silence" threshold added.&nbsp; Keeps VBR 
    from upping the bitrate during very quite passages.&nbsp; (Robert.Hegemann)</font></li>
  <li> <font color="#ff0000">New VBR quality setting from Robert Hegemann.&nbsp; 
    It is based on the idea that distortion at lower bit rates sounds worse than 
    at higher bitrates, and so the allowed distortion (VBR quality setting) is 
    proportional to the bitrate.&nbsp; Because of this, default minimum bitrate 
    is now 32kbs.</font></li>
  <li> <font color="#ff0000">Expermental subblock gain code enabled with -Z.</font></li>
  <li> New "-r" option for raw pcm input files.&nbsp; With -r, LAME will not do 
    any fseek()'s or look for wav and aiff headers on the input file.</li>
  <li> Bug fixes in mp3x (frame analyzer) for viewing frames near end of the file.</li>
  <li> Bug fixed to allow setting the sampling rate of raw pcm input files.</li>
</ul>
<h3> LAME 3.20beta&nbsp; July 19 1999</h3>
<ul>
  <li> Bug in get_audio.c fixed.&nbsp; Libsndfile wrappers would not compile (Miguel 
    Revilla&nbsp; Rodriguez)</li>
  <li> Nils Faerber found some unitialized variables and some wierd extranous 
    computations in filter_subband, now fixed.&nbsp; This was causing seg faults 
    on some machines.</li>
  <br>
  &nbsp; 
</ul>
<h3> LAME 3.19beta&nbsp; July 18 1999</h3>
<ul>
  <li> <font color="#ff0000">Oops!&nbsp; Robert Hegemann immediatly found a bug 
    in the new&nbsp; (old -Z option) quantization code.&nbsp; calc_noise1 was 
    not returning tot_noise, so non ms-stereo frames were buggy.</font></li>
  <br>
  &nbsp; 
</ul>
<h3> LAME 3.18beta&nbsp; July 17 1999</h3>
<ul>
  <li> <font color="#ff0000">Many psycho-acoustic bug fixes.&nbsp; Dan Nelson 
    discovered a bug in MPEG2: For short blocks, the code assumes 42 partition 
    bands.&nbsp; MPEG1 sometimes has less, MPEG2 can have more.&nbsp; In MPEG1, 
    this bug would not have effected the output if your compiler initializes static 
    variables to 0 on creation.&nbsp; In MPEG2 it leads to array out-of-bounds 
    access errors. Finally, there was a related bug in MPEG1/MPEG2, short &amp; 
    long blocks where the energy above 16 kHz was all added to partition band 
    0. (the lowest frequeny partition band!)</font></li>
  <li> <font color="#ff0000">The -Z option (Gabriel Bouvigne's idea of using total 
    quantization noise to choose between two quantizations with the same value 
    of "over") is now the default.&nbsp; I believe this helps remove the trilling 
    sound in Jan's testsignal4.wav.&nbsp; The quality of testsignal2.wav and testsignal4.wav 
    are now better than Xing and getting closer to FhG.</font></li>
  <li> Bug fixes in frame &amp; sample count for downsampling mode. (ben "jacobs")</li>
  <li> Patches to improve modulization.&nbsp; (ben "jacobs")</li>
</ul>
<h3> LAME 3.17beta&nbsp; July 11 1999</h3>
<ul>
  <li> substantial code cleanup towards goal of making LAME more modular.</li>
</ul>
<h3> LAME 3.16beta&nbsp; July 11 1999</h3>
<ul>
  <li> <font color="#ff0000">New tunings of window switching, and better bit allocation 
    based on pe.&nbsp; (Jan Rafaj.&nbsp; improves both testsignal2.wav and testsignal4.wav).</font></li>
  <li> <font color="#ff0000">Bug in mid/side quantization when side channel was 
    zero fixed.&nbsp; (Albert Faber)</font></li>
  <li> Removed some extranous computations in l3psy.c (Robert Hegemann)</li>
  <li> More detailed timing status info, including hours display. (Sakari Ailus) 
    and percentage indicator (Conrad Sanderson).</li>
  <li> <font color="#3366ff">Window_subband and calc_noise1,calc_noise2 speedups.&nbsp; 
    Quantize_xrpow speedup should be significant on non GNU/intel systems. (Mike 
    Cheng)</font></li>
  <li> <font color="#3366ff">Better initial guess for VBR bitrate.&nbsp; Should 
    speed up VBR encoding.&nbsp; (Gabriel Bouvigne)</font></li>
  <li> More advanced .wav header parsing.&nbsp; fixes bugs involving click in 
    first frame. (Robert.Hegemann)</li>
  <li> Correct filesize and total frame computation when using LIBSNDFILE (ben 
    "jacobs")</li>
  <li> Click in last frame (buffering problem) when using libsndfile fixed.</li>
  <li> Audio I/O code overhauled.&nbsp; There is now a uniform audio i/o interface 
    to libsndfile or the LAME built in wav/aiff routines.&nbsp; All audio i/o 
    code localized to get_audio.c.</li>
  <br>
  &nbsp; 
</ul>
<h3> LAME 3.15beta</h3>
<ul>
  <li> times()/clock() problem fixed for non-unix OS.&nbsp; (Ben "Jacobs")</li>
  <li> Fixed uninitialized pe[] when using fast mode.&nbsp; (Ben "Jacobs")</li>
</ul>
<h3> LAME 3.13&nbsp; June 24 1999</h3>
<ul>
  <li> Patches for BeOS from Gertjan van Ratingen.</li>
  <li> Makefile info for OS/2 Warp 4.0&nbsp; (from dink.org).</li>
  <li> Status display now based on wall clock time, not cpu time.</li>
  <li> mem_alloc no longer allocates twice as much memory as needed (Jan Peman).</li>
</ul>
<h3> 3.12pre9</h3>
<ul>
  <li> Updated BLADEDLL code to handle recent changes (Albert Faber).</li>
  <li> Bug fixed in parsing options when not using GTK (Albert Faber).</li>
  <li> <font color="#ff0000">MPEG2 Layer III psycho acoustics now working.</font></li>
  <li> <font color="#3366ff">Improved huffman encoding Chris Matrakidis. (10% 
    faster).&nbsp; I dont know how he finds these improvements!&nbsp; LAME with 
    full quality now encodes faster than real time on my PII 266.</font></li>
  <li> Fixed time display when encoding takes more than 60 minutes.</li>
</ul>
<h3> 3.12pre8</h3>
<ul>
  <li> <font color="#ff0000">New <a href="gpsycho/ms_stereo.html">mid/side stereo</a> 
    criterion.&nbsp; LAME will use mid/side stereo only when the difference between 
    L &amp; R masking thresholds (averaged over all scalefactors) is less then 
    5db.&nbsp; In several test samples it does a very good job mimicking the FhG 
    encoder.</font></li>
  <li> <font color="#ff0000">Bug in mid/side stereo fixed:&nbsp; independent variation 
    of mid &amp; side channel scalefactors disabled.&nbsp; Because of the way 
    outer_loop is currently coded, when encoding mid/side coefficietns using left/right 
    thresholds, you have to vary the scalefactors simultaneously.</font></li>
  <li> <font color="#ff0000">Bug in side/mid energy ratio calculation fixed. (Thanks 
    to Robert Hegemann)</font></li>
  <li> Default mode is stereo (not jstereo) if bitrate is chosen as 192kbs or 
    higher.&nbsp; Tero Auvinen first pointed out that FhG seems to think at 160kbs, 
    their encoder is so good it doesn't need jstereo tricks. Since LAME is not 
    as good as FhG, I am going to claim that 192kbs LAME is so good it doens't 
    need jstereo tricks, and thus it is disabled by default.</li>
  <li> WAV header parsing for big-endian machines, and automatic detection of 
    big-endian machines.&nbsp; (Thanks to&nbsp; Sigbj&oslash;rn Skj&aelig;ret).</li>
  <li> added 56 sample delay to sync LAME with FhG.</li>
  <li> MP3x (frame analyzer) can now handle MPEG2 streams.</li>
</ul>
<h3> 3.12pre7</h3>
<ul>
  <li> MPEG2 layer III now working!&nbsp; lower bit rates (down to 8kbs) and 3 
    more sampling frequencies:&nbsp; 16000, 22050, 24000Hz. Quality is poor - 
    the psy-model does not yet work with these sampling frequencies.</li>
  <li> Fixed "ERROR: outer_loop(): huff_bits &lt; 0." bug when using VBR.</li>
  <li> bash and sh scripts to run LAME on multiple files now included.&nbsp; (from 
    Robert Hegemann and Gerhard Wesp respectively)</li>
  <li> bug fix in encoding times for longer files from&nbsp; (Alvaro Martinez 
    Echevarria)</li>
  <li> yet another segfault in the frame analyzer fixed.</li>
  <li> ISO psy-model/bit allocation routines removed.&nbsp; This allowed makeframe() 
    to be made much simpler, and most of the complicated buffering is now gone. 
    Eventually I would like the encoding engine to be a stand alone library.</li>
</ul>
<h3> 3.12pre6</h3>
<ul>
  <li> <font color="#ff0000">Better VBR tuning.&nbsp; Find minimum bitrate with 
    distortion less than the allows maximum.&nbsp; A minimum bit rate is imposed 
    on frames with short blocks (where the measured distortion can not be trusted).&nbsp;&nbsp; 
    A minimum frame bitrate can be specified with -b, default=64kbs.</font></li>
  <li> <a href="http://www.zip.com.au/%7Eerikd/libsndfile">LIBSNDFILE</a> support.&nbsp; 
    With libsndfile, LAME can encode almost all sound formats.&nbsp; Albert Faber 
    did the work for this, including getting libsndfile running under win32.</li>
  <li> CRC checksum now working!&nbsp;&nbsp; (Thanks to Johannes Overmann )</li>
  <li> frame analyzer will now work with mono .mp3 files</li>
  <li> <font color="#3366ff">more code tweeks from Jan Peman.</font></li>
  <li> <font color="#3366ff">Compaq-Alpha(Linux) fixes and speedups from Nils 
    Faerber.</font></li>
  <li> <font color="#3366ff">Faster bin_search_StepSize from Juha Laukala.</font></li>
  <li> <font color="#3366ff">Faster quantize() from Mike Cheng</font></li>
  <li> <font color="#3366ff">Faster quantize_xrpow() from Chris Matrakidis.&nbsp; 
    xrpow_flag removed since this option is now on by default.</font></li>
  <li> Fixed .wav header parsing from Nils Faerber.</li>
  <li> Xing VBR frame info header code from Albert Faber.&nbsp;&nbsp; "Xing" and 
    "LAME 3.12" embedded in first frame.</li>
  <li> <font color="#ff0000">Bug in VBR bit allocation based on "over" value fixed.</font></li>
</ul>
<h3> LAME 3.11&nbsp; June 3 1999</h3>
<ul>
  <li> Almost all warnings (-Wall) now fixed!&nbsp; (Thanks to Jan Peman)</li>
  <li> More coding improvements from Gabriel Bouvigne and Warren Toomey.</li>
  <li> <font color="#ff0000">VBR&nbsp; (variable bit rate).&nbsp;&nbsp; Increases 
    bit rate for short blocks and for frames where the number of bands containing 
    audible distortion is greater than a given value.&nbsp; Much tuning needs 
    to be done.</font></li>
  <li> Patch to remove all atan() calls from James Droppo.</li>

</ul>
<h3> LAME 3.10 May 30 1999</h3>
<ul>
  <li> <font color="#3366ff">Fast mode (-f)&nbsp; disables psycho-acoustic model 
    for real time encoding on older machines.&nbsp; Thanks to Lauri Ahonen who 
    first sent a patch for this.</font></li>
  <li> <font color="#ff0000">New bit reservoir usage scheme to accommodate the 
    new pre-echo detection formulas.</font></li>
  <li> <font color="#ff0000">Tuning of AWS and ENER_AWS pre-echo formulas by Gabriel 
    Bouvigne and myself.&nbsp; They work great!&nbsp; now on by default.</font></li>
  <li> In jstereo, force blocktypes for left &amp; right channels to be identical.&nbsp; 
    FhG seems to do this.&nbsp; It can be disabled with "-d".</li>
  <li> Patches to compile MP3x under win32 (Thanks to Albert Faber).</li>
  <li> <font color="#3366ff">bin_serach_stepsize limited to a quantizationStepSize 
    of&nbsp; -210 through 45.</font></li>
  <li> <font color="#ff0000">outer_loop()&nbsp; will now vary Mid &amp; Side scalefactors 
    independently.&nbsp; Can lead to better quantizations, but it is slower (twice 
    as many quantizations to look at).&nbsp; Running with "-m f" does not need 
    this and will run at the old speed</font></li>
  <li> <font color="#ff0000">Bug in inner_loop would allow quantizations larger 
    than allowed.&nbsp; (introduced in lame3.04, now fixed.)</font></li>
  <li> Updated HTML documentation from Gabriel Bouvigne.</li>
  <li> Unix&nbsp; man page from William Schelter.</li>
  <li> <font color="#ff0000">numlines[] bug fixed.&nbsp; (Thanks to Rafael Luebbert, 
    MPecker author).</font></li>
  <li> <font color="#3366ff">Quantization speed improvements from Chirs Matrakidis.</font></li>
  <li> <font color="#ff0000">When comparing quantizations with the same number 
    of bands with audible distortion, use the one with the largest scalefactors, 
    not the first one outer_loop happened to find.</font></li>
  <li> Improved defination of best quantization when using -f (fast mode).</li>
  <li> subblock code now working.&nbsp; But no algorithm to choose subblock gains 
    yet.</li>
  <li> Linux now segfaults on floating point exceptions.&nbsp; Should prevent 
    me from releasing binaries that crash on other operating systems.</li>
</ul>
<h3> LAME 3.04 May 22 1999</h3>
<ul>
  <li>Preliminary documentation from Gabriel Bouvigne.</li>
  <li> <font color="#3366ff">I wouldn't have thought it was possible, but now 
    there are even more speed improvements from Chris Matrakidis!&nbsp; Removed 
    one FFT when using joint stereo, and many improvements in loop.c.</font></li>
  <li> "Fake" ms_stereo mode renamed "Force" ms_stereo since it forces mid/side 
    stereo on all frames.&nbsp; For some music this is said to be a problem, but 
    for most music mode is probably better than the default jstereo because it 
    uses specialized mid/side channel masking thresholds.</li>
  <li> Small bugs in Force ms_stereo mode fixed.</li>
  <li> Compaq Alpha fixes from Nathan Slingerland.</li>
  <li> <font color="#ff0000">Some new experimental pre-echo detection formulas 
    in l3psy.c (#ifdef AWS and #ifdef ENER_AWS, both off by default.&nbsp; Thanks 
    to Gabriel Bouvigne and Andre Osterhues)</font></li>
  <li> Several bugs in the syncing of data displayed by mp3x (the frame analyzer) 
    were fixed.</li>
  <li> highq (-h) option added.&nbsp; This turns on things (just one so far) that 
    should sound better but slow down LAME.</li>
</ul>
<h3>LAME 3.03 May 18 1999 </h3>
<ul>
  <li> <font color="#3366ff">Faster (20%) &amp; cleaner FFT (Thanks to Chris Matrakidis&nbsp; 
    http://www.geocities.com/ResearchTriangle/8869/fft_summary.html)</font></li>
  <li> mods so it works with VC++ (Thanks to Gabriel Bouvigne, www.mp3tech.org)</li>
  <li> MP3s marked "original" by default&nbsp; (Thanks to Gabriel Bouvigne, www.mp3tech.org)</li>
  <li> Can now be compiled into a BladeEnc compatible .DLL&nbsp;&nbsp; (Thanks 
    to&nbsp; Albert Faber, CDex author)</li>
  <li> Patches for "silent mode" and stdin/stdout&nbsp; (Thanks to Lars Magne 
    Ingebrigtsen)</li>
  <li> <font color="#ff0000">Fixed rare bug: if a long_block is sandwiched between 
    two short_blocks, it must be changed to a short_block, but the short_block 
    ratios have not been computed in l3psy.c.&nbsp; Now always compute short_block 
    ratios just in case.</font></li>
  <li> <font color="#ff0000">Fixed bug with initial quantize step size when many 
    coefficients are zero.&nbsp; (Thanks to Martin Weghofer).</font></li>
  <li> Bug fixed in MP3x display of audible distortion.</li>
  <li> improved status display (Thanks to Lauri Ahonen).</li>
</ul>
<h3> LAME 3.02 May 12 1999</h3>
<ul>
  <li> <font color="#ff0000">encoder could use ms_stereo even if channel 0 and 
    1 block types were different.&nbsp; (Thanks to Jan Rafaj)</font></li>
  <li> <font color="#ff0000">added -k option to disable the 16 kHz cutoff at 128kbs.&nbsp; 
    This cutoff is never used at higher bitrates. (Thanks to Jan Rafaj)</font></li>
  <li> <font color="#ff0000">modified pe bit allocation formula to make sense 
    at bit rates other than 128kbs.</font></li>
  <li> fixed l3_xmin initialization problem which showed up under FreeBSD.&nbsp; 
    (Thanks to Warren Toomey)</li>
</ul>
<h3><b>LAME 3.01 May 11 1999</b> </h3>
<ul>
  <li> max_name_size increased to 300&nbsp; (Thanks to Mike Oliphant)</li>
  <li> patch to allow seeks on input file (Thanks to Scott Manley)</li>
  <li> fixes for mono modes (Thanks to everyone who pointed this out)</li>
  <li> overflow in calc_noise2 fixed</li>
  <li> bit reservoir overflow when encoding lots of frames with all zeros&nbsp; 
    (Thanks to Jani Frilander)</li>
</ul>
<h3><br>
  <b>LAME 3.0 May 10 1999</b> </h3>
<ul>
  <li><font color="#ff0000">added GPSYCHO (developed by Mark Taylor)</font></li>
  <li> <font color="#000000">added MP3x (developed by Mark Taylor)</font></li>
  <li> LAME now maintained by Mark Taylor</li>
</ul>
<b>November 8 1998</b> 
<ul>
  <li> Version 2.1f released</li>
  <li> 50% faster filter_subband() routine in encode.c contributed by James Droppo</li>
</ul>
<b>November 2 1998</b> 
<ul>
  <li> Version 2.1e released.</li>
  <li> New command line switch <b>-a</b> auto-resamples a stereo input file to 
    mono.</li>
  <li> New command line switch <b>-r</b> resamples from 44.1 kHz to 32 kHz [this 
    switch doesn't work really well. Very tinny sounding output files. Has to 
    do with the way I do the resampling probably]</li>
  <li> Both of these were put into the ISO code in the encode.c file, and are 
    simply different ways of filling the input buffers from a file.</li>
</ul>
<b>October 31 1998</b> 
<ul>
  <li> Version 2.1d released</li>
  <li> Fixed memory alloc in musicin.c (for l3_sb_sample)</li>
  <li> Added new command line switch (-x) to force swapping of byte order</li>
  <li> Cleaned up memory routines in l3psy.c. All the mem_alloc() and free() routines 
    where changed so that it was only done <i>once</i> and not every single time 
    the routine was called.</li>
  <li> Added a compile time switch -DTIMER that includes all timing info. It's 
    a switch for the time being until some other people have tested on their system. 
    Timing code has a tendency to do different things on different platforms.</li>
</ul>
<b>October 18 1998</b> 
<ul>
  <li> Version 2.1b released.</li>
  <li> Fixed up bug: all PCM files were being read as WAV.</li>
  <li> Played with the mem_alloc routine to fix crash under amigaos (just allocating 
    twice as much memory as needed). Might see if we can totally do without this 
    routine. Individual malloc()s where they are needed instead</li>
  <li> Put Jan Peman's quality switch back in. This reduces quality via the '-q&nbsp;<int>' 
    switch. Fun speedup which is mostly harmless if you're not concerned with 
    quality.</int></li>
  <li> Compiling with amiga-gcc works fine</li>
</ul>
<b>October 16 1998</b> 
<ul>
  <li> Version 2.1a released. User input/output has been cleaned up a bit. WAV 
    file reading is there in a very rudimentary sense ie the program will recognize 
    the header and skip it, but not read it. The WAV file is assumed to be 16bit 
    stereo 44.1 kHz.</li>
</ul>
<b>October 6 1998</b> 
<ul>
  <li> Version 2.1 released with all tables now incorporated into the exe. Thanks 
    to <b>Lars Magne Ingebrigtseni</b>(larsi@ifi.uio.no)</li>
</ul>
<b>October 4 1998</b>In response to some concerns about the quality of the encoder, 
I have rebuilt the encoder from scratch and carefully compared output at all stages 
with the output of the unmodified ISO encoder. <a
 href="http://www.uq.net.au/%7Ezzmcheng/lame/download.html"> Version2.0</a> of 
LAME is built from the ISO source code (dist10), and incorporates modifications 
from myself and the 8hz effort. The output file from LAME v2.0 is <i>identical</i> 
to the output of the ISO encoder for my test file.Since I do not have heaps of 
time, I left the ISO AIFF file reader in the code, and did not incorporate a WAV 
file reader.Added section on <a
 href="http://www.uq.net.au/%7Ezzmcheng/lame/quality.html"> quality</a><b> October 
1 1998</b> 
<ul>
  <li> Updated web page and released LAME v1.0</li>
</ul>
<b>Up to September 1998</b> 
<ul>
  Working on the 8hz source code... 
  <ul>
    <li> Patched the <a href="http://www.8hz.com/">8hz</a> source code</li>
    <li> 45% faster than original source (on my freebsd p166).</li>
    <ul>
      <li> m1 - sped up the mdct.c and quantize() functions [MDCTD, MDCTD2, LOOPD]</li>
      <li> m2 - sped up the filter_subband routine using <b>Stephane Tavenard</b> 
        's work from musicin [FILTST]</li>
      <li> m2 - minor cleanup of window_subband [WINDST2]</li>
      <li> m2 - Cleaned up a few bits in l3psy.c. Replaced a sparse matrix multiply 
        with a hand configured unrolling [PSYD]</li>
      <li> m3 - (amiga only) Added in the asm FFT for m68k (based on sources from 
        <b>Henryk Richter</b> and <b>Stephane Tavenard</b>)</li>
      <li> m4 - raw pcm support back in</li>
      <li> m5 - put in a byte-ordering switch for raw PCM reading (just in case)</li>
      <li> m6 - reworked the whole fft.c file. fft now 10-15% faster.</li>
      <li> m7 - totally new fft routine. exploits fact that this is a real-&gt;complex 
        fft. About twice as fast as previous fastest fft (in m6). (C fft routine 
        is faster than the asm one on an m68k!)</li>
      <li> m8</li>
      <ul>
        <li> - Now encodes from stdin. Use '-' as the input filename. Thanks to 
          <b>Brad Threatt</b></li>
        <li> - Worked out that the 1024point FFT only ever uses the first 6 phi 
          values, and the first 465 energy values. Saves a bunch of calculations.</li>
        <li> - Added a speed-up/quality switch. Speed is increased but quality 
          is decreased <i>slightly</i>. My ears are bad enough not to be able 
          to notice the difference in quality at low settings :). Setting '-q 
          1' improves speed by about 10%. '-q 100' improves speed by about 26%. 
          Enoding of my test track goes from 111s (at default '-q 0') to 82s (at 
          -q 100). Thanks to <b> Jan Peman</b> for this tip.</li>
      </ul>
      <li> m9 - fixed an error in l3psy.c. numlines[] is overwritten with incorrect 
        data. Added a new variable numlines_s[] to fix this. Thanks again to <b>Jan 
        Peman</b>.</li>
      <li> m10 - Down to 106 seconds by selecting a few more compiler options. 
        Also added a pow20() function in l3loop.c to speed up (ever so slightly) 
        calls to pow(2.0, x)</li>
      <li> m11</li>
      <ul>
        <li> No speedups. Just cleaned up some bits of the code.</li>
        <li> Changed K&amp;R prototyping to 'normal' format. Thanks to <b>Steffan 
          Haeuser</b> for his help here.</li>
        <li> Changed some C++ style comments to normal C comments in huffman.c</li>
        <li> Removed the #warning from psy_data.h (it was getting annoying!)</li>
        <li> Removed reference in bitstream.c to malloc.h. Is there a system left 
          where malloc.h hasn't been superceded by stdlib.h?</li>
      </ul>
      <li> In Progess:</li>
      <ul>
        <li> my PSYD hack for the spreading functions is only valid for 44.1 kHz 
          - Should really put in a "if freq = 44.1 kHz" switch for it. Someone 
          might want to extend the speedup for 48 and 32 kHz.</li>
        <li> Putting in Jan Peman's quantanf_init speedup.</li>
      </ul>
    </ul>
  </ul>
</ul>
</body>
</html>