Compress MP3 and MP3Pro by 20% losslessly? Options

[Supacon]

This is interesting, but not expressly useful or shocking. MP3 is getting pretty antiquated, so of course one could design an algorithm that compresses superior to MP3, and improves upon its weaknesses. But of course, it's not MP3 anymore, but a new codec.

Probably the only really useful application that springs to mind is for P2P sharing. Perhaps (depending on the licensing and such) players like foobar could incorporate support for this format.

And yeah as Sebastian Mares said, it'd be very nice to have this built-into WinRAR and the like.

[Sebastian Mares]

Hi Sebastian,

Thank you for the interest in the soundslimmer technology. The technology is based on principals of Sound Genome Theory (to obtain preliminary info you may simply Google phrase “Sound Genome Theory” or go directly to http://www.fidelityamplifier.com/Sound_Gen...ory_page_1.html. The theory was developed by a network of scientist which collaboration resulted in many advances in digital audio coding.

If you interested in learning more about principals that SoundSlimmer is based on, we recommend you to evaluate another technology called “SX Fidelity Amplifier” we recently released, that is based on other principals of the Sound Genome Theory. It increases the quality of the decoded signal without increasing the original encoding bitrate. It operates in conjunction with any audio decoding process and provides a “restoration boost” to audio data and delivers a sound output that is similar to the one produced by a higher bitrate signal or a signal with a wider frequency spectrum. If you interested to download the evaluation version, it is located here:
http://www.fidelityamplifier.com/evaluation.html

As you correctly stated Huffman coding is not very efficient, so as a part of the theory advances, we have replaced it with our own proprietary method of coding. In addition to that current evaluation version of SoundSlimmer employs several other technologies including fractal based analysis for the MDCT prediction.

The evaluation version supports MP3 format, and compression of MP3-Pro, AAC, HE-AAC, OGG, WMA, QMF, MPEG-4 audio can be custom built as required. The actual compression ratio depends on the bitrate of the original MP3 file and the actual sound structure of the compressed music.

Both SoundSlimmer and SX Fidelity Amplifier are designed to be highly customizable and can be incorporated into any existing encoding-decoding scheme.

Best Regards,
SoundSlimmer support Team

-----Original Message-----
From: Sebastian Mares
Sent: Sunday, April 30, 2006 3:38 AM
Subject: Technology Used

Hello!

I am just wondering how exactly you are achieving this additional compression. Is it something similar to that what lossless JPEG compressors use, i.e. replacing huffman coding with arithmetic coding and using prediction on MDCT data?

Regards,
Sebastian Mares

This post has been edited by Sebastian Mares: Apr 30 2006, 18:44

[Garf]

That's what you get when the marketing department answers your technical questions

[Eli]

This would be much more interesting for lossless, getting 20% better then flac, wavepack, ape, ect would be a breakthrough.

[Sebastian Mares]

Lossless encoding already uses some of the techniques described here.

This post has been edited by Sebastian Mares: Apr 30 2006, 21:02

[NeoRenegade]

Lossless encoding already uses some of the techniques described here.

Yeah, probably every last one that is applicable.

This post has been edited by NeoRenegade: May 1 2006, 05:23

[sven_Bent]

can the mpz be played diretly form winamp ?

[cabbagerat]

That's what you get when the marketing department answers your technical questions

That's true. But aren't most source coding methods more advanced than Huffman covered by patents (Arithmetic coding, i know, is a field landmined with patents)? Maybe they are just covering themselves because they are using a patented method without licence.

[Destroid]

http://www.rctechnologies.co.uk/

Company has finished development of the first generation of its Sound Restoration Engine called "Radiant SoundT" that rebuilds fidelity of the digital audio. Now any stream that lost its original sound quality can be automatically processed via Radiant SoundT Engine and receive substantial sound quality improvement.

This company called Rainbow Coherent Technologies, LLP that developed SoundSlimmer technology has several interesting projects. Sorry for OT, but if the technology quoted above does what I think it does I know the question about sound restoration has come up in the forums before.

This post has been edited by Destroid: May 1 2006, 20:04

[Eli]

Lossless encoding already uses some of the techniques described here.

Yes, but we dont really know what they are using. Probably as guessed...

[Shade[ST]]

Did anyone try feeding an mp3packer mp3 to this program, to see if it gets any more efficiency or no?

[birdie]

This algorithm/program is wonderful and it has real accomplishments but instead I'd like the author to engage in optimizing lossless audio compression algorithms. First of all no mp3 player producer would use such compression algorithm in foreseeable future and then we all know the inherited weaknesses of MP3 itself so one should get rid of such files completely rather than trying to recompress them even further.

This post has been edited by birdie: May 9 2006, 15:48

[Firon]

Well, it would be useful for distribution on a PC, at least, to save bandwidth (and space if you have players that can play it back)... mpz your collection then send it to a friend, or something. Kind of like raring them.

This post has been edited by Firon: May 2 2006, 07:31

[smack]

Look what for example WinRK, StuffIt, or PAQ can do with JPEGs.
I wonder if soundslimer replaces noisless techniques used in MP3 encoding with its own ones (optimized more for compression ratio than speed) as Garf said, or maybe it is just optimized for squeezing the last bits from huffman-encoded bitstreams. Can someone try to rename some OggVorbis or AAC file to .mp3 and try to feed soundslimer with it?

So, who is going to add specific support for MP3 and Vorbis files to PAQ8?

PAQ8 is available under GNU GPL and already has specific support for JPEG files. Could this method be used in a similar (?) way for the mentioned audio file formats?

[R4Z3R]

Hi you all,

so this is my first post on HA!

in my freetime, iam spending time into programming, mainly into vb6 although it isn't as "hardcore" as delphi/c.
one of my projects was a tool which can analyze the compression rate of files.
so i did a run with both a .mp3 and a mpz. file.... here are the results:

mp3: for me, the hc encoding seems to be relatively good implemented
mpz: as it was said before, ss squezzes really every bit of ballast out of the file, th hc implementaion seems to
XTREMELY good. while there are about 5% (i cant explain the unit, they depend on the structure of my prog )
left of "possible compression", that factor is going to be 0 (!) in the mpz file!

i know, that all are not real news, but i hope it was interessting for those whoare interesstet in the technical backrounds....

PS: my orthography will be better in the next posts

This post has been edited by R4Z3R: May 3 2006, 21:04