KVR Audio

Kingston · Post by **Kingston** » Sun Sep 24, 2006 6:43 pm

the bitrate in MP3 actually just refers to the amount of information in a particular mp3 frame. A psychoacoustics based algorithm decides wether this or that snippet in a given FFT packet (ie, frame) gets included or not. The more data in a frame, the better the sound.

And this is a vastly simplified version.

Lowpass filtering was included in some encoders later on when it was found out that a low-pass filtered signal (ie less sharp transients) would act more gracefully when encoded to very low bitrates.

aMUSEd · Post by **aMUSEd** » Sun Sep 24, 2006 6:45 pm

Use MP3Pro instead?

Warmonger · Post by **Warmonger** » Sun Sep 24, 2006 6:57 pm

The answer "use lossy codec x instead" hardly addresses the question. All lossy formats, by definition, will have their own artifacts. While newer formats may provide better quality at an equal bitrate to MP3, that doesn't help much when you want to distribute the file in a format that anyone can play.

marcanw · Post by **marcanw** » Mon Sep 25, 2006 11:08 pm

I know your problem, I worked on project with music/flash:
PRODIGY : http://www.trust-media.net/prodigy/?ooo=kv
SOLDOUT : http://www.trust-media.net/soldout/?ooo=kv
A low pass is a good start (at around 12 k). It will definitely decrease the size of the mp3. You can mix with the low pass in order to have a coherent mix without the last khz. At the end, don't dither but raher downmix at 32 bit float. Actually, lame (which is known to be the best mp3 encoder) convert everything to 32 bit float before the encoding process. So without dithering you will avoid encoding it (with HF content).

TotcProductions · Post by **TotcProductions** » Mon Sep 25, 2006 11:50 pm

Use 4Front's SandBrush VST. Works wonders.

Mr. Tunes · Post by **Mr. Tunes** » Tue Sep 26, 2006 12:04 am

TotcProductions wrote:Use 4Front's SandBrush VST. Works wonders.

Sorry I'm not sure what you mean by this... do you mean use Sand Brush to emulate what the mp3 encoding process will sound like, or would you use it to restore a mp3 file? The problem i'd find with restoring a mp3 file like that is that after you apply the effect you have to re-encode the file. Curious to know what you meant..

Mr. Tunes · Post by **Mr. Tunes** » Tue Sep 26, 2006 12:08 am

marcanw wrote:I know your problem, I worked on project with music/flash:
PRODIGY : http://www.trust-media.net/prodigy/?ooo=kv
SOLDOUT : http://www.trust-media.net/soldout/?ooo=kv
A low pass is a good start (at around 12 k). It will definitely decrease the size of the mp3. You can mix with the low pass in order to have a coherent mix without the last khz. At the end, don't dither but raher downmix at 32 bit float. Actually, lame (which is known to be the best mp3 encoder) convert everything to 32 bit float before the encoding process. So without dithering you will avoid encoding it (with HF content).

interesting... did you do anything with stereo imaging plugins? just trying to check these sites out now. SolDOUT is giving me errors on the loops

marcanw · Post by **marcanw** » Tue Sep 26, 2006 9:27 am

Mr. Tunes wrote:
marcanw wrote:I know your problem, I worked on project with music/flash:
PRODIGY : http://www.trust-media.net/prodigy/?ooo=kv
SOLDOUT : http://www.trust-media.net/soldout/?ooo=kv
A low pass is a good start (at around 12 k). It will definitely decrease the size of the mp3. You can mix with the low pass in order to have a coherent mix without the last khz. At the end, don't dither but raher downmix at 32 bit float. Actually, lame (which is known to be the best mp3 encoder) convert everything to 32 bit float before the encoding process. So without dithering you will avoid encoding it (with HF content).
interesting... did you do anything with stereo imaging plugins? just trying to check these sites out now. SolDOUT is giving me errors on the loops

I didn't do anything at the stereo image.
For the error on Soldout, it works perfectly for me. Try it again and let me know. Do you have the same problem with Prodigy?
For Soldout, if it works for you, try the sub option, very funny

efdi · Post by **efdi** » Tue Sep 26, 2006 9:55 am

I know nothing about DSP, programing, engineering or whatever.... but here's my modest idea.
Compare the original audio and the encoded audio... Save a footprint or a profile, or something like that with the difference between the 2 files... and then with a specially written program playback the encoded file and process it with the "difference" profile. it might help

or even make an impulse file of a encoding and work with that...
anyway, im just talking outta my.... emm

Krakatau · Post by **Krakatau** » Tue Sep 26, 2006 10:33 am

efdi wrote:I know nothing about DSP, programing, engineering or whatever.... but here's my modest idea.
Compare the original audio and the encoded audio... Save a footprint or a profile, or something like that with the difference between the 2 files... and then with a specially written program playback the encoded file and process it with the "difference" profile. it might help

or even make an impulse file of a encoding and work with that...
anyway, im just talking outta my.... emm

i have't got a clue if it could exist for real but this looks like a smart idea !

marcanw · Post by **marcanw** » Tue Sep 26, 2006 10:42 am

Krakatau wrote:
efdi wrote:I know nothing about DSP, programing, engineering or whatever.... but here's my modest idea.
Compare the original audio and the encoded audio... Save a footprint or a profile, or something like that with the difference between the 2 files... and then with a specially written program playback the encoded file and process it with the "difference" profile. it might help

or even make an impulse file of a encoding and work with that...
anyway, im just talking outta my.... emm
i have't got a clue if it could exist for real but this looks like a smart idea !

It won't work cos the transformation is non linear

Warmonger · Post by **Warmonger** » Tue Sep 26, 2006 1:43 pm

efdi wrote:I know nothing about DSP, programing, engineering or whatever.... but here's my modest idea.
Compare the original audio and the encoded audio... Save a footprint or a profile, or something like that with the difference between the 2 files... and then with a specially written program playback the encoded file and process it with the "difference" profile. it might help

or even make an impulse file of a encoding and work with that...
anyway, im just talking outta my.... emm

That sounds like WavPack

benjamind · Post by **benjamind** » Wed Sep 27, 2006 9:19 am

Best thing is to get the best quality mp3 encoder you can find, LAME is very good and it's FREE. Then encode in the highest quality MP3 bitrate setting which (I think) is 320Kbps stereo (there is an option to use joint-stereo but I don't know what it really does so I'd stick with stereo).

marcanw · Post by **marcanw** » Wed Sep 27, 2006 12:42 pm

benjamind wrote:Best thing is to get the best quality mp3 encoder you can find, LAME is very good and it's FREE. Then encode in the highest quality MP3 bitrate setting which (I think) is 320Kbps stereo (there is an option to use joint-stereo but I don't know what it really does so I'd stick with stereo).

Joint Stereo is definitely better. It just means that rather than encode left and right, joint stereo encodes mid and side.

Lagrange · Post by **Lagrange** » Thu Sep 28, 2006 8:53 am

None has mentioned what impact VBR would have on this process if the effect in question existed.

plugin to fight mp3 encoding artifacts