This is from a message I just posted to uk.rec.audio and rec.audio.opinion:

On Thu, 27 Mar 2003 22:53:58 GMT, Moi <zis_guy@NOxSPAMMxhotmail.com>
wrote:

>This one is from Peter Kater's "Gateway" ca. 1988 (Gaia/Polygram
>Records). This one may in fact exist on CD, so somebody may be able
>to check perceived speed stability and etc. against the CD. The song
>is from side 1, track 2 titled "Talk Back."
>
>This one was also most likely recorded to digital before pressing the
>LP, and I'm not responsible for any possible digital ugliness you may
>hear ;-).
>
>Wish I could post longer samples, but space and downloading time
>constraints prevent it. As again, this one is about 3.5MB in size,
>and requires the LPAC decoder, available here:
>
>http://www.nue.tu-berlin.de/wer/liebchen/lpac.html
>
>The sample itself can be downloaded here:

http://ff123.net/export/pksample.pac

I changed the link to point to my server to remove the bandwidth load
from Moi's site.

Moi described his setup in an earlier message: TD-160/M97xE/Technics
SU-C01.

I purchased the 1988 CD version of this album (< $10 used) and ripped
the same section as Moi's sample.

http://ff123.net/export/pksamplecd.pac

I time aligned it as best I could. The sample is about 34 seconds
long. The vinyl version takes about 60 msec longer to complete than
the CD version, so the speed accuracy of the turntable is about 0.2%
on the slow side if the CD can be considered the reference.

I applied -0.43 dB to the left channel of the CD version, and +0.26 dB
to the right channel. The volume was compared with the vinyl version
via average RMS. WavGain (which uses David Robinson's replaygain
algorithm) gives about the same answer as average RMS.

I applied a linear fadeout to about the last 4 seconds of the sample
to approximate what Moi had done with his sample.

I won't say what I hear just yet. If you think the difference is
sufficiently subtle to warrant the use of a double-blind tool, visit
one of the following sites:

http://pcabx.com/
http://ff123.net/abchr/abchr.html
http://www.kikeg.arrakis.es/winabx/winabx.zip

ff123

Edit: Oops, I think I posted this to the wrong forum

Could you provide a sample of the piano solo in the end of current sample?
Thanks!

I think vinyl sample has much more tonality on the piano than cd: imho cd is sharper, but colder and somewhat less audible.

Since i cannot test the samples (here at work) right now, and since i've listened them with cheap speakers & medium/loud environment, please take this cum grano salis.

I only have the CD, not the LP, so I couldn't provide samples of both to compare unless Moi makes more available. Also, perhaps the end of this sample may be problematic, since I had to guesstimate how Moi did his fade.

ff123

yup...


actually... i don`t need to do a comparison..

i have never heard a digital track... that
"beats" a analouge one..

yes.. am a vinyl-freak..
but to do exact clone is impossible..

it is possible on the other hand.. to do a "perfect"
digital track.. but if so.. the hole reproduction system
must bee 100% digital.. and so on.
but the human ear isn`t compatible....

If you want really good sound I have some wax cylinders in the basement. This is another April Fool's stunt, right? B)

Heheh, that would have been a good one. I should have made some 3.5 MB files available for download which just say "April fools!" on them, one recorded from vinyl and the other from CD.

ff123

Well, there are many levels of comparison:

1. Live feed vs. digital recording
2. analog master vs. digital master
3. direct to vinyl vs. digital master
4. vinyl from analog master vs. CD from digital master

etc., etc.

All of the above would only be fair assuming that the equalization wasn't twiddled with between the different formats. But in real life, this often happens, especially when a recording is re-released and re-mastered. Older vinyl recordings, unless of the "audiophile" variety were often made from a second-generation analog master (or worse).

This comparison is probably:

vinyl from digital master vs. CD from digital master, although the CD doesn't have a code on it, so it could be an analog master. There's no telling how the equalizations differed between the two. But at least both CD and LP were from the same year.

Still, I wouldn't take this comparison too seriously, not knowing all the details. It's just for your entertainment.

ff123

Hehe, most vinyl sound so good,
because they were made only with analog technology.

Mixing digital and analog is the weak point in most mastering studios...
They don't use good DACs/ADCs/filters. (Around here, that is.)
<edit>
Especially AD conversion is messed up. (Look at my flag...)
</edit>

But properly recorded, 100% digitally made CDs beat vinyl...
(means: are more accurate, have better dynamics, are more resistant to wear)
Been there, heard that.

@ff123: Are you sure that this CD is not a remastered version?
<edit>
Heh, I'm writing too slow...
So this is a digital recording.
</edit>

Another question: Will there ever be a Linux version of your abc/HR utility?
I will try to port it with dumb interface (command-line, no ncurses,
ASCII-art, no gpm support(cause I can't code that yet )

I bought the 1988 version of the CD, which was the original release date for both the LP and CD. There is also a 1992 version of the CD, which I didn't purchase.



Carsten Haase was working on a Linux version when he was interrupted by a birth in the family. He says he has a workable version, though. You might wish to contact him.

ff123

I am actually using some tool written by the author of MAD.
It's very generic, takes two commandlines and mixes them,
asks which is which few times and after the test gives out the probability.

(Sorry, can't find that guy on the web... email, webpage?)

There was a Perl script written by Monty (Ogg Vorbis):

http://ff123.net/128test/abx_perl1.txt

There is LinABX:
http://www.beryllium.net/~remco/linabx/

Frank Klemm also wrote a command line version for *nix, which is in the lame source code as abx.c, but that version is uninteresting

The author of MAD is Rob Leslie.

ff123

Most Vinyl tracks I have heard sound superior than that of CD. I do not believe this means vinyl is technically better. My Vinyl rips recorded at 24 bit, dithered to 16 bit, sound better than original CD's. I believe this is because sound engineers master CD for todays tiny stereos, radio, car and portable players. The CD format is not weak, its the usage of unnatural dynamics compression and eq by engineers that make vinyl sound better.

And what about this version ?

piosample.pac ?

@ff123: Thanks, these utils are great, but none shows probability...
and none can compare multiple files. That's why I'll make my own product...
(wxWindows 2.4.0, platform-neutral, statically linked versions will be avaliable... command line later)
The problem is, that these are LARGE (1MB at least...
Sorry, ld (GNU linker) includes too much of wxWindows code even with minimal samples.
Luckily they seem to not grow much with increasing complicity.)

<edit>
@Pio: The server is very slow... That version will download whole day
</edit>

No other webspace lets me upload files bigger than 1 MB. It's not the server itself, I just downloaded it in less than 60 seconds (very max speed of my 512 kbps line). It must be the ISP connection with the web, too bad...
If someone with a better server sends me his mail by PM, I can mail it.

Well, I've downloaded all of them... that server wasn't all that horrible...
Why have you posted these in LPAC format, not FLAC?
(I have to dload decoder now )

The original poster of the LP sample (Moi) used Lpac, so I kept things the same.

I updated piosample.pac.

I realized that it was beginning with a short silence. So if you want to perform ABX tests with the other files, either remove manually the silence at the beginning, either redownload it.

CD sample has much more high frequencies, sounds like major quantization problems.
(or vinyl is loosing much of high freqs)
Pio's is so similar to 'original' but has very slightly less muffled sound.

Test result (Pio vs. 'original') 7 of 8, p = 0.035, Pio's sample has 'cleaner' plates at ~4 sec.

<edit>
I've removed the silence already, thanks.
</edit>

Hehe, I see what you did, Pio. I was thinking of doing something similar myself. This version sounds quite close to pksample.pac, and really does warrant the use of ABX. It's not quiet in my house right now, but I'll try to give it a go later.

ff123

I don't know alot about the tech. side of audio, but just lestioning with my raw ears(which really it all comes down to) I have always thought that vinyl sounds significantly better than cd's.



Life would be alot better if I could marry an actress and my cd's would sound as good as vinyls.



-Darin
:alien:

I definitely hear the higher frequencies in the CD version much better. The two samples are definitely different enough not to need ABX, in my opinion.

However, the vinyl is nowhere near unlistenable (somebody had a good clean pressing too). Since I've been listening mostly to vinyl the past couple years, CD's have begun to sound uncomfortably "chiseled" to me (maybe due to the wider frequency extension, possibly to some other factors). The vinyl sounds "mellower" and easier on my ears, although somewhat less detailed and 'receded'. It may just be a matter of getting/being used to listening to one rather than the other.

It should be noted perhaps that we're hearing a CD (digital) vs. "vinyl recorded to digital" (lord knows what ADC was used) -- so it isn't a true comparison of vinyl vs. CD. Interesting exercise nonetheless.

Edit -- sounds like the frequency balance of the vinyl is pretty good, being well under 1dB difference in channels compared to the CD version (assuming the same balance was used on both the vinyl and CD recordings... seems logical).

Agreed, for the pksamplecd.pac file. I can hear this difference pretty clearly even through my labtec computer speakers

piosample.pac is pretty close to the vinyl version though. I believe Pio did some equalizing to the CD version to make it sound similar.

ff123

P.S. Check your private messages.

Pio, you're good. I don't believe I can tell the difference between your sample and the vinyl version.

ff123

Have a fun... You can find cases when a LP sound better than CD (the same interpret and title) but from technical point of view is LP inferior to CD, so where's the problem? I used to be a vinyl freak (as I own more than 5000 LPs) but I would never switch back to LP now as I know how good a CD may sound when properly done.

Would you like to donate some of those LP's to a "good cause" ?

P.S... surely you must admit, there's a big catalog of music that was never released on CD... so why not listen to both? IMHO it's an excellent reason to listen to vinyl if you like that older stuff (which are often closer to the master tapes than 'remastered' CD's), and records are cheap too. I've discovered a lot of "lost" music, plenty of gems that the RIAA never saw profit in re-releasing on CD.

I wonder if the vinyl version could be EQ'ed to sound nearly identical to the CD (i.e. the reverse of what Pio did)? Maybe there's a project for you, ff123.

For a good (almost April fool's) vinyl vs CD comparison...

If you have the Beatles CD "A Hard Day's Night" - the album, compare the track "You can't do that" from the CD with youcantdothat.wav from the lame samples site:
http://lame.sourceforge.net/download/samples/

I don't think you'll need to ABX And no, it's not a different song.


So, for occasions like this, you need vinyl.

Cheers,
David.

P.S. Sometimes though the CD version of an old recording beats the LP. e.g. Frank Sinatra - Come Dance with Me - the CD isn't a great CD, but it's good, and the LP is terrible. This is quite rare though - usually the first release of a recording is the one which gets the most care, so LPs from the 1950s-1980s are often worth a listen.

Damned !
My process has been guessed !

I pasted two pictures of the spectrums of the samples, and applied a graphic equalization matching the difference.

The equalized CD sample is still sharper than the vinyl one, because there are more dynamics in the treble.
RMS level difference between cd and piosample is <1db, and the peak level is 1 db inferior on the vinyl.

After a -35 db high pass filtering, 2500 to 5000 Hz, so as to keep only treble, the RMS levels are still within 1 db, thanks to the equalisation, but the vinyl peaks have fallen 5 db below the equalized CD ones !
That's why the vinyl sounds smooth.

I should now try to apply a multiband dynamics compression on the equalized sample in order to try to smooth completely the treble.

I recall for those who don't browse the hardware section, that I recently performed a real analog vs digital blind test at home with hardware switching.
http://www.hydrogenaudio.org/forums/index....ST&f=21&t=7953&

I Think (because I'm not positive) that the mastering for vinyl incorporates EQing, or sound processing if you prefer, not to make it sound better, but, rather, to compensate for the medium. Whereas, for CDs, the engineer does not have to worry about the sound of the stylus scraping across the vinyl or tape being dragged across the playback heads, and the final result is truer sound. As a result, vinyl may sound better because of the processing, while CDs give a more accurate representation of the original recorded sound. Dolby is another example of sound processing--EQing--that further compensates for the limitatons of tape.


Later.

I don't own music simultaneously on LPs and CDs; the older releases I have on LPs and newer on CDs... but the CDs go back only to early 90's, when they really started to sound reasonably. A lot of CDs I got are the golden ones from DCC and MFSL (from the original masters) and JVC's jazz XRCDs. Anytime I get the possibility to listen to a CD-title I have on LP I listen to it and then I decide if I keep on having the LP or change it for CD... So I have a lot of remasters or Japan-imported disks, which sound really good. A lot of doesn't, though.

About donating - although I am definitely not a Salvation Army member, if I'm going to do something like this, then I get you on the top of my list - I mean it...

You are wrong... When mastering a CD, the engineer must worry a lot, because he knows, the format reveals everything done wrong or inproperly... When mastering a LP-material, you must worry about the limits of the vinyl (or the transferring medium, i.e. stylus), especially in low frequencies. It's just different attitude and it's not easy to switch from one to another as it may seem...

Hey, I appreciate it . It might be a bit expensive to ship from Europe to the USA, but if some were LP's I've been looking for (perhaps hard to find in the USA), it would be worth it.

Could you explain this further? When you say 2500 to 5000 Hz, it sounds like you used a band-pass filter. If high-pass, above which frequency did you keep -- 5KHz?

Anyway, I think it's clear there's significantly less treble in the vinyl version. Whether that has to do with the cartridge that was used or the vinyl, is up in the air (although I regularly hear more treble with CD's myself, a more expensive and/or better cartridge might do better in the treble). Ff123 says the cartridge was a (Shure) M97xE, which I think is MSRP $140 and about $75 street price (not expensive, but it should do the trick)... hmmm.

I once borrowed Horowitz playing Rachmaninoff's 3rd piano concerto on vinyl from the library. The record was very thick (it was from 1951 and they used lots of material), but it was old and scratched -- just terrible condition. But it must have been the first pressing, because all the high frequencies were there, nonetheless.

I later bought one of those re-releases on LP because I liked that rendition so much, and was horrified by just how much loss of high frequencies there were in that version.

So I went to a record collectors store and bought an earlier pressing from the 1950's, but damned if it still wasn't bad.

The CD release of it wasn't too bad, except they applied that stupid pseudo-stereo stuff to it, and I still think the library record sounded the best, even with all the scratches! I am appalled to think of all the great performances sitting around on first-pressing records which I'll never hear.

ff123

Actually, you could hear some of them... finding them would be the trick. There are lists and such (favorite pressings of certain records by various people) altho I don't have any specific links.

At the price of records at some places ($0.25 to $0.50 apiece) it's worth the gamble if something looks promising. That's basically a price where it wouldn't hurt just to chuck it in the trash if it sounded bad. Classical LP's in particular you can often find near-mint at throwaway prices.

I've kept an article in french, by a sound engeener of the times of vinyl, telling about the art of processing master tapes so that the vinyl pressings sounds like them. According to him, a record factory could refuse a master tape arguing that the sound of it could not be properly cut on a vinyl.



2.5 to 5 kHz is the transition band. If I'm not mistaken, it means that the 5 kHz is kept, and that the 2.5 kHz is a the attenuation I set (-35 db).



Here are the spectrums, from 1 to 22 kHz.



Blue is the CD, grey is the vinyl, and red the equalized CD. The vertical lines are spaced by 11 db



The vinyl can play a bigger role than the cartridge, go back in http://www.hydrogenaudio.org/forums/index....f=1&t=2896&st=0 (samples available on request). On the same record, the treble of track 2 of side 1 was 2 db below the CD, and the one of track 4 side 2 was 9 db below the CD. !

I tried this again, this time focusing on the impact of the snare drums, and I could tell the difference this time (ABX 16/16). Interesting that when you know what to listen for, it gets easier to hear.

ff123

Good morning! I say (write) it here from the very first moment I discovered this forum... But always got laughed at... Mostly I don't really need any ABXing... The only thing you must have is EARS

And what do you think ff123 used to hear the differences?

The only difference is that his procedure is reliable, whilst sighted ones are not guaranteed to be.

(Edit: removed some extra irony)

I typically use ABX when I'm in doubt that I can hear a difference, as in this case. An added benefit is that other people who might also have trouble hearing a difference can trust my results (assuming I report high confidence ABX numbers).

There are some listeners whose opinion I trust, and don't really need to see ABX results to believe them. Others I'm not so sure of. But ABX removes the doubt. And to reiterate KikeG's point, I only use my ears to form an opinion.

ff123

I definitely need ABX, even for myself. I can play the same track over and over without hearing the same thing twice. Thus I always think I hear a difference, even between A and A.

The same happens to me, mostly deep in the night, when I am really tired. But hearing the same piece again under another conditions usually removes all doubts aside.

@ budgie. Yes, you basically said the same thing I did with one exception. I said "the engineer does not have to worry about the sound of the stylus scraping across the vinyl or tape being dragged across the playback heads." I did not say that they, the engineers, do not have to worry. And, you are quite correct; CDs do demand care and attention by the recording engineers, as every little defect will show itself.

Later.

Hey Pio..

What's your audio editor over there?

Just curious..

Jon

Not sure if your'e talking about RIAA equalization or something else.. You HAVE to
use the RIAA eq, but because it's a standard, and is accurately (hopefully!) re-compensated
for in your phono preamp, it's transparent and should result in no change to the sound.

Since you are carving little sound waves in the record, but want to keep the speed of the needle
motion to a fairly consistant maximum, you have to (if I remember right) emphasize the low frequencies
nearly 20 db, (cause they're slow) and de emphasize the highs a similar amount. (cause they're fast)

This is why you can't just amplify your cartridge and put it into your soundcard or whatever.

Tape as I recall is much more linear, (though I'm not as sure of this) and dosen't need quite
as drastic of a eq as the really huge one that is the RIAA standard.

All this should be independent of the "sound".. just ways to make the medium "flatter" and
more transparent to the listener. What the sound engineer does to affect the "sound" is another
story..

I wasn't talking about any specific standard. I read an article (about 12 years ago) on the differences of vinyl/tape vs CDs, and that is all I remember. Not much, I know, that's why I said "I'm not positive."

Later.

That's why I desperately buy every "original masters" CD/LP I'm interested in as I see it, regardless of the price... and the price is high

Spectrum analysis : SoundForge 4.5 (+spectrum plugin ?)
Overlay of the curves : Corel Photopaint 9
Graphic Eq : Shibatch Super eq + Winamp discwriter.
Parametric eq (highpass) + RMS levels : SoundForge 4.5