listen to this.

edit: link down

Can you provide the original uncompressed sample? Sounds like clipping in the CD master to me.

i don't have the original anymore , it was a lent CD from a friend.

In this case we can't get anything from this sample.

Well, can you at least tell us the name of the album and song? Perhaps somebody else might own it and be able to provide a reference sample.

Please PM me when you have information that can make this at least semi-relevant.

edit: re-opened.

artist: system of a down
cd title: system of a down
song title: know

I have that CD, so I can provide my samples.

Here is the original losslessly compressed sample.

Here is the MP3 I got when I encoded it with LAME 3.96 --aps.

Note that this is quite a loud song (-10.6 dB track gain).

Edit: My ISP's webserver seems to be acting very weird. Sometimes I can't connect, sometimes I get a 404 when trying the link, sometimes I get a random picture (!?!), and sometimes it just works. I also tried the attach the files to my post put apparently I can't attach files larger than 2MB here.

Please keep retrying, hopefully it will work eventually.

Edit2: OK, it seems to work now. /me knocks on wood...

is it normal to get a "know-sample.jpg" ?

btw, how does lame --preset fast standard behave on it?

No, it's not. But as you can see in the first edit of my post I had these problems as well in the beginning. Right now, everything seems to work fine. Can you try the links again? If you still have this problem, try clearing your browser's cache.


Check this attached MP3 file...

The encoded > decoded sample provided by audioflex sounds worlds apart from the sample I generated by encoding the flac (using Lame 3.96 --preset standard) provided by PoisonDan. No need for ABX here.

I'd surprised if audioflex's sample was actually encoded using aps or the wav wasn't processed in some way prior to encoding or the CD it came from was an mp3/lossy generated copy.

Yup, my thoughts exactly.

@audioflex:
Can you tell a bit more about how you created your MP3 file? Did you rip straight to 3.96 --alt-preset standard or has there been any other encoding or processing done?

It was straight off the cd (used cdex for ripping), which leaves me wondering why poisondan's copy sounds worlds better.

You should perhaps extract your track to wav and listen to the wav file. Perhaps the problem lies in the extraction.

edit: too bad you mentionned that you do not have the cd anymore.

Was it a "real" (ie pressed) cd or a cdr?

real, pressed.

Well, Since PoisonDan's encode seems to sound OK, I think we can rule out LAME as the problem.

Tip for ABXing, concentrate on the background symbols.

perhaps it was a bug in the ripper *shrugs*, all the other songs that i ripped off the cd sound fine.

edit: typos.

there were different versions of this cd one with explicit lyrics, one without. Perhaps there were other versions as well and they were mastered diffferently?

Kristian

audioflex,

Could you post the actual MP3 part, instead of the MP3 converted to APE?

I guess you decoded it to WAV first so you could use a standard audio editor to cut that part out and encode to APE.

However, there are also MP3 editors that don't re-encode the file, like e.g. mp3DirectCut. You could use this on your original MP3 to create the MP3 sample.

I'd like to analyze the actual MP3 file in EncSpot, to check if there are any obvious differences with my MP3.

Edit: Would it be too much trouble to ask your friend if you can borrow his CD again?

though this says nothing of the quality of course, the spectrogram shows a lot of lowpassing at 11025. like it's been transcoded _several_ times between minidisc recorders.

that would explain the awful sound too.

if this was a pressed CD, please find out who pressed it and report them to the relevant authorities

we don't want people charging full price for this rot, do we?


that is of course, provided LAME isn't making similar artefacts for some unknown reason (modifying the source and b0rking something), or assuming the sample wasn't transcoded _after_ being ripped.

although spectrograms are meaningless for determining sound quality (in cases where they are meaningful, they don't show anything the ear can't perceive, and are thus not useful), they will tell you immediately if something is from a lossy source, and provide some helpful information that the untuned ear might miss (or the untrained brain may be unable to interpret, as is my case)

audioflex,

I'd also be interested to know what you get when you use aucdtect on your MP3 file.

Edit: Download here.