MP3 or MiniDisc?

[OtherWeis]

I received this Joe Jackson show in a CDR trade with a dude from Germany over a decade ago. A couple of the discs proved to be things that still have not popped up on these sites (I upped the Pre-FM Joe Jackson from the Old Waldorf 79 that I got in that trade). This show from The Beacon Theatre in New York 2001 is an audience recording, and I have never analyzed one of those before. Looking at it, to me it is clearly lossy, however, the frequency drop off is not as sharp as I would expect from an MP3. I know that, at that time, minidisc was a frequently used, but imperfect, device for capturing audio. From the Spectrogram and Frequency Analysis below, would you say that this is lossy because of MP3 in the lineage, or because it was recorded on a device that compressed it at the start? Is there even a way to tell?

[tonebloke]

MD.

[paddington]

frequency roll off is a poor indicator of lossy data compression. It might be from a codec or it might be due to the source. It doesn't mean anything, by itself.

Plenty MP3 modes don't roll off, at all.

You must look for unnatural sounds. Those big holes and hard, deep steps are clear indications of lossy data compression. The audio is missing in places the codec believes you won't notice. And you generally don't... what you *do* notice is the aliasing from the edges of the missing chucks. That's the glassy / slushy sound you'll hear in lossy digital audio.

That doesn't look like a clear case of ATRAC (MD) compression to me, though it certinaly could be a later version of ATRAC which looked more like MP3. Or maybe it's just a higher-bitrate MP3.

Either way, it seems to have been through a lossy compression codec at least once.


the most glaring ATRAC compression and most common before 2008 or so looks like this:

[paddington]

however... and this is really important:


it looks like you are using Audacity for spectrum analyzer output. It's about the worst there is for discerning whether something is lossy.

If you really want to know, take a 3 sec snap shot, from Adobe Audition (or Cool Edit Pro 2.x), in color, and post that. It's a far superior representation of the what's actually there.

[OtherWeis]

Unfortunately, I do not have Adobe Audition. Is there anything else I can use (i.e. free)?
This is an excellent recording of an excellent show, and I really think that it should be heard.

[Neil Wilkes]

The easiest way to spot MP3 sources (and once you know what to listen for, lossy sources in general) is tio take the suspect file, and open it up in your DAW/Editor and instead of playing it in regular mode pop across to voxengo.com and gran the free MSED plugin.
Plonk this onto your track, and mute the Mid component & now pan the S component to both speakers.
Listen.
Ideally, just to learn what you are listening for, make a low bitrate MP3 (at 128kbps) and then import the MP3 and the source file (from CD or wherever) and do the Mid-Si8de thing on the master buss. Toggling between the MP3 and the lossless will clearly reveal the chirping sound you get from MP3/lossy sources. Learn this sound & become allergic to it in the same way you would to the sound of out-of-phase material.

Once you can hear this in worst case scenarios (128kbps MP3 is about as bad as it gets) you will start to find you can spot this sound even in regular listening modes.....
Seriously - this is a much better way to tell than looking at spectral plots. Sources taken from old vinyl can punk out at 15k - look for the label "ffss" on vinyl. This means "full frequency stereophonic spectrum" and was common in 60's & even the 70's, and your upper limit was around 14k400.

[Audioarchivist]

Originally Posted by OtherWeis

Unfortunately, I do not have Adobe Audition. Is there anything else I can use (i.e. free)?
This is an excellent recording of an excellent show, and I really think that it should be heard.

I think that the free trial version of Adobe Audition (i.e.: won't let you save, etc.) will still allow you to view spectral graphs you can screencap...

Furthermore, at a certain point recently, Adobe ran into some kind of glitch that basically allowed people to download full working "free" copies of almost all their CS2 line, so, you could get a free working copy of Audition (on the honour system, I guess... haha). I don't have the info handy, though, anymore... But, google around a bit - you might find it!

[conan1982]

Howdy,

This may be my recording from April 20, 2001 at The Beacon Theater; if so, I used a minidisc recorder to originally capture the audio. I was sitting in the first row, orchestra left for this, and the December 2000 show, which I recorded, as well. At the time I had fairly recently switched over from analog cassette, was not aware that minidisc was a lossy source.

Would it be possible that you could up this show here so that I could give it a listen to see if it is, in fact, mine? I am especially interested as I can only find the original second and third minidiscs for this and I would like to have a friend work on a new edit of this gig.

Thanks in advance.

Conan1982

Originally Posted by OtherWeis

I received this Joe Jackson show in a CDR trade with a dude from Germany over a decade ago. A couple of the discs proved to be things that still have not popped up on these sites (I upped the Pre-FM Joe Jackson from the Old Waldorf 79 that I got in that trade). This show from The Beacon Theatre in New York 2001 is an audience recording, and I have never analyzed one of those before. Looking at it, to me it is clearly lossy, however, the frequency drop off is not as sharp as I would expect from an MP3. I know that, at that time, minidisc was a frequently used, but imperfect, device for capturing audio. From the Spectrogram and Frequency Analysis below, would you say that this is lossy because of MP3 in the lineage, or because it was recorded on a device that compressed it at the start? Is there even a way to tell?