I know so little about compression technology...I can hear differences -sometimes big time - when I transform a wav into an MP3 but I'm not quite sure what is being referred to when they say it is "lossy." What is being lost?
Even more important -for someone relying upon MP3 hosting sites - is there anything that can be done to polish and burnish and mp3 to have it "gain back" some of what was lost? EQ? Some sort of dynamic exciter?
I know about some of the other compression technologies coming down the pike (or already on the pike, as it were) but it seems as if Mp3's remain the standard for the sites that I post on. I'm tired of hearing stuff that I've slaved over become pale and wan doppelgangers after undergoing the transformation, though.
Any thoughts? Advice?
Thanks.
MP3 Therapy
-
- KVRAF
- 1927 posts since 30 Oct, 2003 from Frolicking in Dirac's Ocean
-
- KVRAF
- 3588 posts since 13 May, 2004 from montreal
Try an a/b comparison of your original .wav with the .mp3 version. Should answer your question about what's getting lost.
-
- KVRist
- 162 posts since 15 Apr, 2003 from new zealand
obviously the more you compress it the more you lose. try using high quality VBR or at least 256kbps CBR mp3s. if you can get the right encoder, you can fiddle with the complex settings like total bandwidth, high and low cutoffs.
the standard high frequency cutoff of mp3s seem to be 16kHz. but you can tweak your encoder to push this up a bit, as it gets rid of quite a bit of the "air" and such.
the standard high frequency cutoff of mp3s seem to be 16kHz. but you can tweak your encoder to push this up a bit, as it gets rid of quite a bit of the "air" and such.
-
- The Teach
- 8273 posts since 23 Jul, 2002 from flatness
rob-
- KVRAF
- Topic Starter
- 1927 posts since 30 Oct, 2003 from Frolicking in Dirac's Ocean
wow-thanks, all...pHz -the "teach" - most apt...precisely what I was looking for - whether I can understand it or not is a different questions. Thanks!
saved me some effort there -
- KVRAF
- 12235 posts since 18 Aug, 2003
In many cases, I don't find mp3s at higher encoding rates too badly damaged. Anything less than 192kbps though and it can really get bad. I've never been a big fan of VBR, as I find sometimes my ear doesn't adjust as well to the lossed frequencies when it constantly changes. At times, it would be like listening to mix between FM and AM radio.
I'm afraid that once those frequencies have been removed from the file, there's really no way to get them back. Aural exciters and the lot might help in a few instances, but really they would either do nothing or just add unpleasant distortion.
One thing I have often wondered is whether mp3 is kinder to certain sound recordings than others and what that threshold is? For instance, would a dry snare sound fare better under compression than a wet reverb snare sound?
I'm afraid that once those frequencies have been removed from the file, there's really no way to get them back. Aural exciters and the lot might help in a few instances, but really they would either do nothing or just add unpleasant distortion.
One thing I have often wondered is whether mp3 is kinder to certain sound recordings than others and what that threshold is? For instance, would a dry snare sound fare better under compression than a wet reverb snare sound?
-
- The Teach
- 8273 posts since 23 Jul, 2002 from flatness
good question ...
... presumably the less frequencies a sound contains the better it compresses ???
so your dry snare might well compress better than the reverby one ???
(unless the few frequencies it might comprise are all near or above the .mp3 cutoff)
slainte
rob
... presumably the less frequencies a sound contains the better it compresses ???
so your dry snare might well compress better than the reverby one ???
(unless the few frequencies it might comprise are all near or above the .mp3 cutoff)
slainte
rob-
- KVRAF
- 12235 posts since 18 Aug, 2003
There have been times when encoding some of my tracks that I couldn't really notice a difference compared to the original, but in most cases I've never really sussed the pattern to what made it better in those cases (confess to never really being so concerned that I spent more than 10 minutes wondering).
Shame that so many online delivery services still stick with 128kbps. The degree of damage to audio can be quite maddening at that rate.
Shame that so many online delivery services still stick with 128kbps. The degree of damage to audio can be quite maddening at that rate.
-
- KVRAF
- Topic Starter
- 1927 posts since 30 Oct, 2003 from Frolicking in Dirac's Ocean
I think the MP3 process really knocked the shit outta the last piece I posted (Xibalba) - the "compelling" nature of the symptohm sounds (which is why I posted it) just seemed to have some of the life force/chi sucked outta them. The piece was very reliant upon the interplay between complex harmonics that I heard (or that I heard) in the real highs and lows
-
- KVRer
- 7 posts since 5 Apr, 2005
this is a good point and example of what MP3 compression does to a full range audio file...pHz wrote:
... presumably the less frequencies a sound contains the better it compresses ???
so your dry snare might well compress better than the reverby one ???
Now first of all, you have to give the creators of the format some credit... its been around for years and still stands up as a credible format, also given the extreme amount of reduction you get in the file size, on what is arguably the most complex of data, it still amazes me that this format (and all the new fangled ones) are even listenable,
But unfortunately you can't smash a 30 meg file to bits and expect it to come back the same.... One of the major things that is always noticable on an MP3 is the lack of depth... It almost goes completely 2 dimensional, even though the source might be a fantastic (live) analogue recording... this is'nt qantifiable in terms of frequencies or EQ ( remember most 70's vinyl recordings tail off very sharply at 16KHZ ) it's just the nature of the beast.... the only parallel example I can think of is the difference between an acoustic 192KHZ 24 bit file and and 44.1KHZ 16 bit file recorded from the same analogue source... Reverb tails, and complex sounds and harmonics are just mashed by the process of audio compression
-
- KVRAF
- 3508 posts since 27 Dec, 2002 from North East England
Excellent SoS article, but be aware that it's a little out of date in it's recomendations.
'Alt preset standard' is generally regarded as 'transparent' and usually gives an average bitrate (it's VBR algorithm) of around 190.
If you need (or want) to stick with a constant bitrate, 'Alt preset CBR' will have you sorted.
I still haven't found a program which makes better sounding MP3s than CDex + the presets, but I'm open to suggestions.
Stay the hell away from Xing for a start. LAME is the way you want to go. CDex (google it) includes LAME without needing to download anything extra. The presets (based on hundreds of listening tests) are excellent, and should provide the best tradeoff between sound quality and filesize, regardless of bitrate .SoS wrote:A good encoder such as Xing Technology's Audio Catalyst (available from www.xingtech.com/mp3/audiocatalyst/)...
'Alt preset standard' is generally regarded as 'transparent' and usually gives an average bitrate (it's VBR algorithm) of around 190.
If you need (or want) to stick with a constant bitrate, 'Alt preset CBR' will have you sorted.
I still haven't found a program which makes better sounding MP3s than CDex + the presets, but I'm open to suggestions.
-
- KVRAF
- 3508 posts since 27 Dec, 2002 from North East England
Another thing. To me, low bitrate files hinge on whether they employ joint stereo encoding or not (CDex presets always do). At 128 it's the difference between a sloshy mess and a decent listen. Some will tell you that joint stereo f**ks up the stereo spectrum, which is true of older codecs, but it's not an issue in LAME anymore (certainly not to me). Try both and see what you think.
-
- KVRian
- 809 posts since 25 Apr, 2004 from Windsor, Ontario
artists and musicians should really fight back and adopt another form of lossless audio compression.
Last edited by CC4 on Thu Apr 07, 2005 12:10 am, edited 1 time in total.