mp3-sourced or just a low quality recording/transfer?

[Gizby]

The show I have is Nirvana 9-28-1991, and the lineage I have is fairly bland [ANA(2) > CDR(5) > EAC > WAV > FLAC > WAV > CDR] (back from when I would burn all shows to CDR and would as soon grab a master transfer as a CDR(5)). Someone is interested in getting this show from me. When I ripped it, I decided to FLAC it just to see what kind of compression it would get (level 5). The overall compression ratio is 0.3072. Seems kind of low to me. I've attached an analysis of the left channel for SLTS in Anal Freq and about a 1-second spectral view of the same song in EAC. Any idea if this has been mp3'd at some point in the past? Thanks.

[Five]

that's pretty lousy, but I can't conclude it's mp3-sourced from that.

pm me I'd like to get a sample of one of the tracks and take a look & listen (about 5-10 seconds is enough).

Also maybe we could get you to do SA with another prog
http://www.thetradersden.org/forums/...ead.php?t=6436

[Lou]

Could be noise reduced too, not that that's good either and, as far as I'm concerned just as bad as mp3. Digital noise reduction is terrible.

[oldbrokentapes]

I'm pretty sure the only way a FLAC ratio can get that low is from the file being mono, or the overall RMS volume being very quiet. I'd presume the former is the case here.

yeah i'd say poor recording (although if it had been converted to mp3 you probably wouldn't be able to tell because it's so bad already lol)

[Gizby]

I suspect you're right, rob. Should have picked up on that from the start. Thanks.

[Five]

got the sample from Gizby. This show is 99.999999% mono. The channels are identical except that the right channel peaks 00001 higher, as reflected in the stats:

Left Right
Min Sample Value: -28763 -28764
Max Sample Value: 28414 28415
Peak Amplitude: -1.08 dB -1.08 dB
Possibly Clipped: 0 0
DC Offset: -.526 -.526
Minimum RMS Power: -40.28 dB -40.28 dB
Maximum RMS Power: -9.67 dB -9.67 dB
Average RMS Power: -15.17 dB -15.17 dB
Total RMS Power: -14.7 dB -14.7 dB
Actual Bit Depth: 16 Bits 16 Bits

Using RMS Window of 50 ms

Also worthy of note is that the dc offset could use correcting. When I corrected it on the sample, I got a little lift in the high frequencies.

This doesn't sound lossy whatsoever and I can't find any blocks when I do a SA zoomed in on 2 seconds. So my theory is that this is a completely lossless mono source that has been either over nr'ed or perhaps recorded with a $2.99 microphone. The sound is not that bad, considering.

[ssamadhi97]

Originally Posted by Five

got the sample from Gizby. This show is 99.999999% mono.
[...]
This doesn't sound lossy whatsoever and I can't find any blocks when I do a SA zoomed in on 2 seconds. So my theory is that this is a completely lossless mono

I'd agree on that (but not the NR theory). Probably just a not very impressive mic or high analogue generation.

By the way, in Audition you can try using the channel mixer with the "vocal cut" or "LR to Mid-Side" presets. That way you can see the difference between the stereo channels (the side channel). Very easy to tell whether the track is mono or not from there

[ssamadhi97]

Originally Posted by Five

Also worthy of note is that the dc offset could use correcting. When I corrected it on the sample, I got a little lift in the high frequencies.

erk wtf? that definitely shouldn't be happening.

dc offset correction does not affect any frequencies above 0Hz (dc, doh )

[Five]

That's what I'm seeing in the FA. I can send you the clip and steps to follow to duplicate the result on your machine.

[ssamadhi97]

Yea, I'd be very interested in that. Because in theory this is complete nonsense. So I'm wondering wtf is going on there.

[Five]

I know very little of the theory... I just go by experience. I'm far from a mathematician (did I spell that right? lol).

Here's screenshots of CEP FA with the entire 10-second sample selected and at two arbitrary points (74221 and 357412 using the sample display).



I used effects > amplitude > amplify preset called "center wave" (pictured).



Then I took 3 screenshots, one with everything selected and the other two at the exact same specific points. Take the before and after FA shots into an image editor (you'll have to switch from index to rgb mode) and line them up and click between the layers. I'm seeing the highs lift.



Check your pm for a yousendit.com link. If anybody else wants this sample file, send me a pm.