Does anyone know tools for restoration of sound encoded at low bitraits ?
Let's say mp3 has terrible quality..
So software should consider that artefacts have digital source, not analog..

There are many programs for restoration sound from vinyl and tapes..
But how about this ? Anyone heard something?

No.

Any attempt to fix said issues would result in more issues of a similar nature. You can't rebuild a unique wall if many of the unique bricks were already destroyed.

it help alot to think logcally

IF you could do it
Then all playback devicse would just do it to get better quality.
and then it would be the same as playbakc and ther would be no possibiltiy to make it better then normal playback.

Now you have a logically paradox..
So NO you can't


Try seen if anything has at logicall paradox is an easy way to get a 99% secure answer


it goes for the sayin

recompression 128kbits to 320kbits enhacens quality = paradox = NO
making wav before burningenhances quality = paradox = NO
etc etc etc


This only works under the requriments that everything works and is not fbuggy

"The paradox is only a conflict between reality and your feeling of what reality ought to be." -Richard Feynman

Do you mean making a wav out of audio cd before burning or do you mean mp3->wav?

How 'bout a good tool to uncompress heavily dynamic compresssed modern music

Make it more living, breathing, with crisp clear sound, and not every instrument trying to top each other in loudness, resulting in a horrible flat, noisy, boring experience....

I think what the original poster might be looking for is something to "fake it". It could never restore the original sound, but it might make it less unpleasant to listen to. As an analogy, if you rip a CD that is badly scratched and get errors, you still get a better result if you interpolate the missing samples (usually not noticeable audibly) instead of leaving them as arbitrary data (a click). Running a noise removal filter to remove tape hiss from a recording doesn't restore the sounds that the tape hiss destroyed, but it still can make it more pleasant to listen to.

Something you might be able to do to lossy compressed files would be to take some of the lower frequencies and duplicaet them to try to recreate some upper harmonics that were lost in the lowpass. I don't know how well it would work, but it's worth a try. AFAIK this is the principle that SBR works on. There are also tools for deblocking and/or deringing overcompressed JPEG files (some video codecs actually implement deblocking and deringing as a optional post-processing feature in the decoder). The result loses some detail, but makes some of the artifacts less ugly.

I don't think there's any software to do this for mp3 files though. There have been threads on this before, perhaps if you find those there might be a solution.

Exactly Phong.
I asked about tools that act more like postprocessing filters for sound.
In order to improve sound just a little bit.. till acceptable quality level for listening.
Sure it would be far from original..

But it seems i've lost my hope for that
Thanks guys anyway.

If you just want to do some experiments with dynamic range expansion to see what you can achieve, it can certainly be done with CoolEdit/Adobe Audition. Just use the Transform|Amplitude|Dynamics Processing feature, where you can draw a graph of the required input-to-output level transform. You can even listen to it while you fiddle with the graph to see how it affects the sound.

I'm not an expert with Audacity, but as far as I can see there isn't any way to do it using that program. There is a compressor function, but it doesn't seem to be able to do expansion. (If anyone can correct me on this point I would be grateful).

EDIT: Sorry, I got sidetracked by Loke's comment about heavily compressed music, and forgot that Xeone's original query was about low bitrate MP3s. Obviously my posting is only relevant to dynamically compressed music.

Theorecticly it could be done but...

You would have to know the compression ratio it was compressed at, the attack time, the decay time, hard or soft knee compression, single or multiband compression, etc.

Unless you knew all the parameters of the compression that was applied, you would just end up with pumping or slurring of the audio.

Just use flac and you won't have to worry about it.

While you're quite right if you want perfection, even without knowing all the compression parameters you can usually make a fair stab at it. But it takes time, a lot of fiddling with the settings on your dynamics processor, and a patient ear. That said, the "UNcompressor" default in the Waves plug-in "Linear Multi-band" (aka LinMB) is rarely bad straight off. I find radio is the hardest to uncompress since the compression is especially hard. Over compressed mastering can usually be fixed a bit.

Restoring over-compressed MP3 (low-bit rate) files is currently not especially successful, but I am also encouraged by the recent work on filters for JPEG artifacts, so perhaps a filter will be made to reduce birdies and other MP3 problems. I have had some success with modest use of MaxxBass to add a bit of low end drive, and various Exciter software to push the upper treble. The exciters are not especially successful though since the birdies only get accentuated as well. More development is needed here.

The technology does exist though. I've worked out a number of coded audio improviments. As with any dynamicly program adjusted parameter, it can only better the smarter the control mechanisms are. As par for the course with these kinds of things, detecting that the live source material actually has a particular coding issue is more than half the battle.

Often times some improviments designed for "de-codefying" audio and video can have usually desireable effects even on lossless audio, depending on how 'smrt' your processing is. imo and experience a robust ai network is needed to have perfect balance of all parts of the processing.

unfortunately this is "a little" more difficult than loading a few directx/vst plugins.

[edit] i've used this to restore missing frequencies sometimes. i wouldn't recommend it for use in a broadcast or batch-processing setup. http://www.yohng.com/mb.html#sandbrush [/edit]