MPC (audio compression format)
|
|
MPC, or 'Musepack', is an open source optimized version of the MPEG-1 Audio Layer 2 / MP2 lossy audio compression format.
Unlike MPEG-1 Layer II, it uses:
- subband-wise selectable M/S encoding (like AAC)
- Huffman coding (like MPEG Layer III, AAC, ...)
- noise substitution techniques as used in ATSC A-52 and MPEG-4 AAC V2
- pure variable bitrate between 3 and 1300 kbit/s (when needed)
The psychoacoustic model of MPC is based on MPEG ISO model 2, but is extended by CVD (clear voice detection). The quantization algorithm of the MPC encoder performs spectral shaping of the noise, called adaptive noise shaping (ANS), in order to overcome the low frequency resolution of the polyphase quadrature filter bands.
In May 2004, a series of double-blind listening tests [1] (http://www.rjamorim.com/test/multiformat128/results.html) (as reported on Slashdot [2] (http://slashdot.org/articles/04/05/24/0623247.shtml?tid=141&tid=185&tid=188)) suggested that Musepack and Ogg Vorbis 1.1 (which was the "aoTuV" fork at the time) would be the two best available lossy codec for high-quality audio compression at bitrates around 128kbit/s, compared with MP3, AAC, WMA, or ATRAC. Other listening tests have also rated MPC highly. However, most of those following tests are in medium/low bitrate area, for which mpc hasn't been designed. And so they aren't really informative on musepack goal, which is to obtain 100% transparency at --standard setting (~160-180kbits average).
Listening tests of MPC:
- 2004 - rjamorim's second 128 kbit/s group listening test [3] (http://www.rjamorim.com/test/multiformat128/results.html) - between 14 and 27 listeners. MPC and Vorbis tied for first.
- 2003 - rjamorim's 128 kbit/s group listening test [4] (http://www.rjamorim.com/test/128extension/results.html) - between 14 and 29 listeners. AAC, MPC, Vorbis, and WMA tied for first.
- 2002 - ff123's second 128 kbit/s group listening test [5] (http://ff123.net/128test/instruct.html)
- 2001 - ff123's 128 kbit/s group listening test [6] (http://ff123.net/128tests.html)
- 2004 - High bitrate tests [7] (http://www.hydrogenaudio.org/forums/index.php?showtopic=18397) (thread by a Musepack developer)
Development of MPC was initiated in 1997 by Andree Buschmann and later taken over by Frank Klemm, and is currently maintained by the Musepack Development Team (MDT) with assistance from Frank Klemm.
Encoders and Decoders are available for Microsoft Windows, Linux and Mac OS X free of charge, mostly licensed as free software under the LGPL or BSD license. They are distributed via the MPC project's website (see external link below). Many third-party programs (audio players, CD rippers etc.) support MPC, either directly or via plugins.
MPC uses the APEv2 tag metadata container.
In the past, MPC has been under suspicion of violating multiple patents (MP2, PNS, subband) [8] (http://www.hydrogenaudio.org/forums/index.php?showtopic=6473). According to the developers of MPC, all patented code has been removed and it is now believed to be free of patent encumbrance. However, one PNS patent application is still active [9] (http://appft1.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PG01&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.html&r=1&f=G&l=50&s1=%2220050004791%22.PGNR.&OS=DN/20050004791&RS=DN/20050004791), and it is not trivial to know if MPC's own "noise substitution techniques" avoids its scope or not. Only a careful examination by a patent lawyer would say whether Musepack is truly patentless or not.
External links
- Musepack.net (http://www.musepack.net) Official website of the Musepack project (encoders, decoders, XMMS, Beep Media Player, Winamp plugins, forums)
- Frank Klemm's MPC page (http://xmixahlx.dyndns.org/audio/files/linux/musepack/klemm/www.personal.uni-jena.de/~pfk/mpp/) (mirror)
- Musepack forum (http://www.hydrogenaudio.org/forums/index.php?showforum=56) at Hydrogenaudio.org
- Pymusepack (http://www.sacredchao.net/quodlibet) Python bindings for libmpcdec (BSD decoder library). Used by Quod Libetde:Musepack