Confusing LP2 spectrum

[oldbrokentapes]

Just to throw a spanner in the works of range_hood's nice guide, the following confuses me:

Taped dredg last night, with Sony ECM-717 (frequency response 100-15,000kHz) > Sony MZ-N710 (LP2)

Looks like


Sounds like
http://www.megaupload.com/?d=18222G3M

Answers on a postcard...

[Five]

oh wow

I'll probably get a chance to look at this tomorrow

[Five]

this doesn't resemble any of the sources range_hood has documented in the ATRAC Thread. I'm wondering which version of ATRAC you MD uses. I couldn't find your model listed here or here. Probably ATRAC ? (>= 4.0) fwiw.

this one's quite tricky to spot. I'd love to see a before/after. It isn't showing the usual drop in the frequency response whatsoever, so the FA doesn't show anything that I can see. However, if you zoom to a 2-second window using spectral view you can see the telltale rectangles and squares punched like holes. Your deck supposedly only goes up to 15kHz but in reality it reaches much higher. Maybe it's more accurate up to 15kHz.

btw, was this a direct digital capture or did you go in thru the analog?

okay, here come the pics.

CEP:

[Five]

Audacity:

[Five]

Eac:

[Five]

oh yeah...

are there any other quality settings (LP4 etc) available on your deck? I'd like to check out something recorded using another setting if at all possible.

[range_hood]

As far as my researches go, your Sony MZ-N710 has got the Atrac DSP Type S codec. Where wikipedia states the only difference to Type R is an advanced LP codec.
Still waiting for your sample to finish download (3k/s). Really looking forward to.

Could you please do some samples to update our guide with authentic analyses?
Maybe just 20 secs. I´d recommend http://www.yousendit.com/ to send large files.

[ssamadhi97]

yea transfer details would be interesting

ps: sources like this = nail in aucdtect/tau analyzer's coffin

Originally Posted by range_hood

I´d recommend http://www.yousendit.com/ to send large files.

either that or http://www.rapidshare.de/ (unlimited dls and bandwidth, size < 50MB, iirc)

[range_hood]

Originally Posted by ssamadhi97

yea transfer details would be interesting

tripled

Any normalization in there?

[oldbrokentapes]

Sorry for the slow reply - I'm... lazy

Anyway, firstly, I taped NIN on Friday night using the same setup and it's in a way reassuring to see this look for more "normal"



I think that most of my tapes do follow more typical patterns. When I was listening to the dredg recording, the highs seemed a lot better than they usually did, which prompted me to check out the spectrals and find such a nice appearance.

The transfers are currently analogue (digital ones will be done in the future, at which point this thread can perhaps get bumped) - I've made the following to show the level of noise floor that is being worked with. The short section at the beginning is with the MD unit on pause; thereafter it's playing back a section where the mic was not turned on. Obviously there is some noise, but by analogue transfer standards I think this cheap soundcard actually performs pretty well (a FA plot show basically a straight line along -90dB).



If I recall correctly, I had the levels set spot on when transferring so no normalisation was required.

And yes, apologies for the slowness of megaupload - I've been giving it a try as rapidshare is no good for people with transparent-proxy-ISPs and yousendit's expiration and download limits began to annoy me.

[Ted]

Not to change the subject, but I'm trying to learn here. What's the purpose of seeing the noise floor? Is it to give an indication of how "noisey" the recording unit is? Would you make a profile (or whatever it is called) of that noise and then subtract it from the recorded signal to get a more accurate version of the recorded signal? Or am I wrong on both counts? No lengthy tutorial or anything, just a quick reply will suffice (thanks).

[oldbrokentapes]

In this case, it's just to give an indication of how much, erm, "more black" the frequencies at the top of the spectrum would/will probably appear given a cleaner transfer method

[Ted]

The answer is "None. None more black." (Spinal Tap, if ya didn't know)

Ok, thanks. I get it now.

[range_hood]

robkismet, could you please change the resolution for future spactrals?

Options > Settings > Spectral Tab > "256" Bands

[Five]

I'm totally baffled how the dredg show has that nice response in the highs It would be fantastic if you could get that on all yer tapes.

some of the fuzz is noise, but that doesn't explain the transients. perhaps hotter levels can cause pleasant distortion in the highs (on a limb here...).