Hey,

I got this APE file on my computer, and wanted to check if it really was lossless. So I decoded it and made a graph in EAC, here's what i got:



It seems to me there is a lowpass around 19.5 kHz....
Does this mean the APE file was transcoded from a lossy source, or is that not to be automatically concluded.
Is it for example possible that the original compact disc material was already lowpassed?

Thanks in advance for your help

It seems to be from a lossy source, especially looking towards the beginning and end of the file where the sound isn't as complex and the lowpass is higher. I'm not a programming or psychoacoustic expert or anything, but it really does look like a lossy source.

Simple answer: Mokeys Audio is lossless!

I have a lot of original CDs which have been lowpassed at 18 or 19 KHz.

I think you misunderstood me. Off course Monkey's Audio is lossless, but the source file which was encoded with Monkey's Audio could possibly not be the original CD, but a lossy file like an mp3.

Maybe the topic title wasn't totally adequate

The source was definately lossly: There are "holes" for a lack of a better term in the spectrum around ~8 kHz after the 10 second mark where the encoder left out some of the audio because it was relatively quiet and would have been masked by the rest of the sound present.

Looks like a lossy source. Here's my spectrum straight from the WavPack:

Thanks, you both have convinced me.

Too bad though I transferred 300megs of decompressed mp3 or such

Hmm, same track and same artist i presume...?

I think. Norah Jones - Feels Like Home - 01 - Sunrise (my guess, at least, from the title of the EAC window)

I use auCDtect.exe to detect if a file is from the original source:
ftp://tta.iszf.irk.ru/aucdtect-lastest.zip

Also with the lowpass being higher in parts where the sound should be less complex (ie. the beginning and end of the file) I'd assume its from a CBR lossy source.

Sorry, I misunderstood this part of your post (which imo is very misunderstandable)!

Don't trust P2P, buy the CD.

...and hope that the CD doesn't suffer from other flaws such as low dynamic range and the like. But yes, CDs indeed do win.

It's a very unreliable program and I wouldn't trust it too much. I've read many posts elsewhere where people have tested it extensively and got a lot of false negatives or false positives.

I think that identifying between a lossy and lossless source is an action the human brain will always be superior in comparison with any computer algorithm. Well, that is, providing there is some aid, such as a spectrum analysis.

yup

i agree

i should

i agree, although i needed some help

In my experience it is very reliable if you analyse a whole cd.
You can group a track by track rip using -d.

I disagree. I've posted some examples somewhere here in this forum...one problem I had was that the original wav file got a probability of 90% whether the lossly LAME encoded file got 95%

So it is not just that it identified a lossy LAME file as "CD" but also that the original file got a lower rating than the lossy file...

Cant that be from a MD recording?

I'd say that's most definitely an MP3 encoding; the behavior of the spectrogram at and above 16kHz is a giveaway for MP3's so-called "sfb21 limitation." Look at the spectra around 16kHz - there's an obvious difference in the amount of information just above and below that frequency. Below, there is a lot of signal/noise present, and above, there is just a few spikes of what may be louder, more sudden attacks. This looks much like what happens when one uses the -Y switch on LAME, or feeds a signal to an MP3 encoder that doesn't lowpass at bitrates like 192kbps.

Just an educated guess, of course.

From my past experience with MD, no. Before I got rid of my last MD recorder last year, I remember studying some of the signals in a spectrogram. For one thing, the most recent version of ATRAC doesn't lowpass at all - some signals are present past 21kHz, in fact. Plus, ATRAC is much more "spikey" looking, with lots of holes going all the way to the lower frequencies. Again, the smooth lowpass at 16kHz with occasional spikes of HF material makes me think that we are looking at an MP3 encoding, but without the original file that was (probably) used to make the first poster's lossless file, it's just speculation.