Why Are Downloaded MP3s Such Trash?
Nov 25, 2002 at 11:02 PM Thread Starter Post #1 of 23

kerelybonto

doo-di-doo-di-dum doo-di-dum doo-di-doo-di-dum
Joined
May 6, 2002
Posts
1,518
Likes
10
When I feel like listening to plasticky, hollow music, I fire up my Philips computer speakers and let my collection of about 1400 downloaded MP3s (most of which I also own on CD) roll. I'm in a mood for this masochism more than you may think, since I'm often moving around doing multiple things that don't allow for me to be tethered to my SACD player via my HP-2s.

I can't listen to my extensive compressed music collection on my headphones. It's too painful. Now, I understand that the people who originally rip the music don't use the best software for the job and that there's inevitably some error introduced when a file's passed among users thousands of times. But still, some things I just don't understand. Like the mysterious treble spike. I've found quite a few tracks that, even over my computer speakers, make me wince as soon as they come on -- there's some high treble tone embedded in the MP3 file that is just annoying as hell, much like the sound a television emits. So I whipped out Cool Edit and took a look:

attachment.php


attachment.php


Frequency response graphs from two separate files show what's basically a 16kHz tone sitting in the file. Both of these files were encoded with Fraunhofer IIS.

How does this kind of thing happen? Is it on the encoding end or does it happen in transmission? If the latter, will (legal) digital music transmission via the internet ever be viable, or will these sorts of errors always pop up?

Grr. Anyway.

kerely
 
Nov 25, 2002 at 11:11 PM Post #3 of 23
Yeah, I know what you mean. Try some lower-priced audiophile headphones like the SR-60 or the KSC-35. They seem to work very well with MP3s. Also, it certainly is a ripping/encoding error, since as far as I know transmission over the Internet is bit-perfect.
 
Nov 26, 2002 at 12:01 AM Post #4 of 23
That's what I had thought -- but it seems really hard to believe when you consider all the digital goop that ends up in these MP3s. All the chirps and things I thought were more likely from transmission.

I guess, in the long run, it's better that they're not. Now if people would just start ripping correctly. ...

kerely
 
Nov 26, 2002 at 2:08 AM Post #5 of 23
The problem is that people don't rip correctly. If everyone did LAME + EAC, we would have no problems. Oh, I also have a few friends that convert all of their mp3s down to 128, to save space. That really helps quality. . . :-/

Does anyone remember the bladenc(I believe that's what it was) encoder? Every time I would encode something, I'd get warbles in the treble.
rolleyes.gif
I mean, just blattant distortion of some sorts. Sort of sounded like ET was trying to phone home. otoh, lame was perfect. . .
 
Nov 26, 2002 at 2:14 AM Post #6 of 23
I noticed a lot of bad 128 mp3's have a constant noise in the background, is this the treble spike you are talking of? The worst is when I downloaded about 40 .wavs off kazaa and found out that 80% were just bad mp3's.
frown.gif


I can't trust mp3 format anymore.

Biggie.
 
Nov 26, 2002 at 2:24 AM Post #7 of 23
The biggest problem with mp3's that you download isn't the common HF roll-off, it is the widely available godawful mp3 encoders like Xing and Blade, and people who transcode between different formats. But hey, one can't complain about something s/he gets for free, right? Or can he..
 
Nov 26, 2002 at 3:02 AM Post #8 of 23
you're getting your mp3's from the wrong sources...

KaZzA, Morpheus, WinMX, etc. etc... dont cut it

most ripping "groups" online rip high quality rips with LAME at 192 kbps which is like mp3 standard

vandit
 
Nov 26, 2002 at 3:46 AM Post #10 of 23
I still find it hard to believe that all of these errors are due to the codecs. If they're that bad, why do they still exist? -- is it because the good ones, like LAME, are usually command-line and don't integrate into playback apps?

So have any of you experience this 16kHz spike I mentioned? You'd know it if you heard it -- it's not at all like background noise, Biggie, it's like someone's torturing a rabbit inside your speakers (or headphones) -- well, for those sensitive to high-frequency sounds, I guess ... I had a friend over and he couldn't hear what I was talking about.

kerely
 
Nov 26, 2002 at 9:40 PM Post #13 of 23
I really only listen to MP3s I cut myself from my own library of original CDs. Of course, I use EAC+LAME. I must say, however, that I would I would NOT recommend this combination to most of my friends because they are unlikely to be able to figure them out and use them reliably. They just don't have the attention span, or spare time to spend learning how to use these tools. Especially if you want to calibrate EAC to use the correct read and write offsets for your specific CD drive. It is quite possible to create imperfect and bad sounding MP3s using EAC and LAME if you don't know what you are doing. Most people need a drop-dead simple program with limited options. That is why the crappy commercial mp3 encoders still predominate.

The few MP3s I have downloaded off the net, (LEGALLY!), I keep in a separate directory from my main library because I don't want to pollute it with the inferior quality "free" mp3s.

I loaned an MP3 disc I created of the complete Beethoven symphonies to some of my friends here at work. These people have been saturated with crappy "free" MP3s downloaded from the net. Each of them independently remarked on how good the Beethoven disc sounded and that they couldn't believe they were MP3s.


FWIW, Zurg
 
Nov 27, 2002 at 12:02 AM Post #14 of 23
LAME 3.93 just came out but do not use it because it seriously degrades sound quality (compared to older versions). Refer to www.hydrogenaudio.org and their mp3 forums for more details.

I was using 3.92 before but I recently reverted to 3.90.2 because it is 10% faster and encoding (and recommended).

With the program EncSpot, you can actually see what type of encoder was used to encode your mp3 files.

it's available here: http://www.guerillasoft.com/EncSpot2/

I scanned a whole bunch of mp3s I downloaded from the web and surprise surprise, the majority were encoded using Xing or FHG fast encoders.
 
Nov 27, 2002 at 12:39 AM Post #15 of 23
Quote:

Originally posted by jessica00
well, i just took a comparison, musicmatch @ 192 vs EAC+LAME at 160, lame sounded better of course..over HD-497's nonetheless
for those wondering., song was sasha - xpander.


someone has good taste
smily_headphones1.gif
 

Users who are viewing this thread

Back
Top