Strange FAs

[tilomagnet]

First I want to say that this is my first post in this section and thanks in advance to everyone who can help me out. I have been checking shows I received in trades for mp3 sources for years already, but last week I got two shows in I'm absolutely not sure about whether they are lossless or not. I did a frequency analysis on them as usual and first it looked quite ok, as there were no cuts in the spectral line or anything like that. But then I noticed that the spectral line from 0 to 11 khz and from 11 to 22 khz look almost identical. :hmm: (Sorry it's hard for me to explain it better, but I think it becomes obvious when looking at the screenshots.)

I have never seen such a FA before and as I got them both from the same trader I suspected that maybe they had been mp3 sourced and someone manipulated the FA somehow to make them look like lossless. :confused: I know this sounds strange and I can hardly imagine how that could be done, but would that be possible? Is anyone able to tell me if these are "true" lossless sources which I can trade? Both are audience recordings in not so great quality, but they are the only sources available, one is from '78 the other one from '81. Thanks in advance for any help.

Here are the screenshots(I will post SAs as well if necessairy):

http://www.geocities.com/tilomagnet/1978_01.JPG

1978

http://www.geocities.com/tilomagnet/1981_01.JPG

1981

[range_hood]

(I will post SAs as well if necessairy):
Yes please.

[ffooky]

Without a doubt this has been (poorly) upsampled from 22 to 44.1KHz.

[tilomagnet]

Here are the SAs of the 1981 show:

http://www.geocities.com/tilomagnet/SA_1981_01.JPG


http://www.geocities.com/tilomagnet/SA_1981_02.JPG



Without a doubt this has been (poorly) upsampled from 22 to 44.1KHz.Thanks. So are these 'former' mp3 sources without any doubt?

[Rider]

Here are the SAs of the 1981 show:

http://www.geocities.com/tilomagnet/SA_1981_01.JPG


http://www.geocities.com/tilomagnet/SA_1981_02.JPG



Thanks. So are these 'former' mp3 sources without any doubt?

Nope that's not what that means at all.

[tilomagnet]

What does it mean then? Are these tradeable or not? :wtf:

[Five]

can you post sa zoomed to 2 seconds, please?

[spontabmark]

am i the only one that hates the B&W spectrals? :lol

[tilomagnet]

Is that the correct one?

http://www.geocities.com/tilomagnet/SA_1981_04.JPG

[Five]

yeah, that's great.

can we see two more in colour: one zoomed to 2 seconds and one thats a couple minutes. I'm wondering if this will help somebody figure out the strangeness. Doesn't look lossy that I can tell so far.

[tilomagnet]

http://www.geocities.com/tilomagnet/SA_1981_05.JPG

http://www.geocities.com/tilomagnet/SA_1981_06.JPG


Thanks for your help so far. But I'm wondering if this isn't lossy, how can you make a FA look like that? If you transfer something from a tape I don't think it will ever have such a FA. ffooky mentioned upsampling, but from what frequency originally? And why would someone want to upsample this?

[range_hood]

I stumbled upon this by accident, when showing my mom some spectrals and had just EAC to create samples.

Exact Audio Copys "Decompress" feature was used. It "converts" to 44100 Hz by default.

http://img143.imageshack.us/img143/6972/mp35622khzeac5cf.png

http://img307.imageshack.us/img307/4462/mp39622khzeac7gv.png

[calendar]

I have a spectral analysis graph like tilomagnet's that is from a mp3 that was recorded as it played through a websites flash audio player

[calendar]

http://zincpaint.tripod.com/graphz.html

[ssamadhi97]

Without a doubt this has been (poorly) upsampled from 22 to 44.1KHz.
Indeed. This mirror image stuff is the infamous aliasing, and the frequency the spectrum is mirrored at is the original sampling frequency / 2. In most cases in this thread the spectrum is mirrored at 11025Hz, so the original sampling frequency was 22050Hz. You can observe the same effect with upsampled 32kHz material, where things get mirrored at 16kHz.

Good resampling filters will apply a not too aggressive lowpass / roll-off around original sampling frequency / 2 or do other tricks to prevent aliasing, which always results in a more or less faint mirroring effect around this frequency in the spectrum of upsampled material.

The characteristic that visually sets resampled material apart from the cutoff introduced by mp3 and other lossy encoders is the steepness of the lowpassing effect and the lack of any mirroring - with lossy encoders, the higher frequencies are usually not lowpassed away but simply *not encoded* which results in a very steep cutoff.