PDA

View Full Version : Comparing 128k AAC to 256k AAC?



decryption
12th January 2009, 12:08 AM
Anyone know how to tell if a 128k AAC file has been transcoded to be 256k AAC? I have a suspicion that the "upgraded" iTunes Plus files are just re-encoded from 128k AAC to 256k AAC, not re-encoded from the original CD source.

NeoRicen
12th January 2009, 12:12 AM
They'd be in a lot of trouble if that were true. Although wouldn't they sound worse if that was the case? I'm sure plenty of people would have noticed that.

decryption
12th January 2009, 12:14 AM
They'd be in a lot of trouble if that were true. Although wouldn't they sound worse if that was the case? I'm sure plenty of people would have noticed that.

My main concern is how the hell did they get all the publishers to re-rip and re-upload their non iTunes Plus content without telling them why? Plus I'm listening to some tracks I've upgraded and they sound horrible, worse than the 128k versions.

I want more proof until I go stirring shit though, hence my request for a method to scientifically find out rather than my shit ears.

Back2Bedlam
12th January 2009, 12:14 AM
Considering it just a level of compression, it's likely, that you can re-encode and regain that amount of data

ford.boy
12th January 2009, 12:14 AM
It would sound slightly worse. I'm sure some audiophiles around hear could tell.

Phase
12th January 2009, 12:16 AM
You can almost guarantee masters were taken and copied, I remember when iTunes first launched they talked about how all their music was copied by Apple using high end equipment, on site, etc etc.

I wouldn't worry about it, you're not getting an 'upscaled' track.

decryption
12th January 2009, 12:18 AM
You can almost guarantee masters were taken and copied, I remember when iTunes first launched they talked about how all their music was copied by Apple using high end equipment, on site, etc etc.

I wouldn't worry about it, you're not getting an 'upscaled' track.

That's not how it's done now though - now it's just the label that rips the CD using a special application Apple gives out to the label and it adds it to Apple's collection of files.

Phase
12th January 2009, 12:23 AM
That's not how it's done now though - now it's just the label that rips the CD using a special application Apple gives out to the label and it adds it to Apple's collection of files.

When you're done crushing my hopes and dreams, remember to close the door, or the cat will get out.


:(

decryption
12th January 2009, 12:24 AM
When you're done crushing my hopes and dreams, remember to close the door, or the cat will get out.


:(

Also, Steve Jobs' mother wears army boots. SALT IN YOUR WOUNDS.

The Drummer
12th January 2009, 12:24 AM
They'd sound worse if they were.

I don't think I have any to compare and it's not like you can easily export them both to a lossless format for others to hear... not without going to audio CD first anyway.

decryption
12th January 2009, 12:27 AM
They'd sound worse if they were.

I don't think I have any to compare and it's not like you can easily export them both to a lossless format for others to hear... not without going to audio CD first anyway.

I have a CD, a 128k AAC and an upgraded 256k AAC iTunes Plus file of the same album - is there a way I can tell from these?

Phase
12th January 2009, 12:29 AM
Anthony makes the transition from consumer into paranoid audiophile, loses friends and becomes 'that guy'.

News at 11.

sjobs
12th January 2009, 12:31 AM
Apple would never deceive their customers. Apple is 100% honest about everything they do.

Right?..

The Drummer
12th January 2009, 12:35 AM
I have a CD, a 128k AAC and an upgraded 256k AAC iTunes Plus file of the same album - is there a way I can tell from these?

Any way you can take the same excerpt from all 3 versions and upload them as a AIFF or WAV or something?
It only needs to be 10secs or so.
Choose a part with a fair bit going on in all frequencies like a chorus or something.

decryption
12th January 2009, 12:36 AM
Any way you can take the same excerpt from all 3 versions and upload them as a AIFF or WAV or something?
It only needs to be 10secs or so.
Choose a part with a fair bit going on in all frequencies like a chorus or something.

Yeah, I can do that :)
Will need to be tomorrow though, I need to find the CD.

Also, how would I convert the DRM protected track to AIFF/WAV?

Beau
12th January 2009, 04:31 AM
Unshackle your iTunes purchases! | Beau Giles (http://beaugiles.net/blog/2008/12/unshackle-your-itunes-purchases/)

Will just strip out the DRM without re-encoding. :thumbup:

EDIT: Or iMovie '06 will let you import the DRMed files, and then export it as an AIFF file. How stripping the DRM was done ages ago.

fompsweeva
12th January 2009, 04:55 AM
Strip out the DRM and use a spectral analyser to few the frequency spectrum the song covers.

Audio/sound analysis OS X and Macintosh applications (http://www.macscience.net/MacScience/Audio_analysis.html)

There are some tools there.

Of course a listening test could be considered science, if done by multiple people. Maybe you could do a blind listening test by posting a wav of each sample rate, and people vote which one they think sounds better.

If it's re-encoded, the 128k is likely to sound better due to less artifacts left over from the re-encoding.

stevejay
12th January 2009, 06:50 AM
You need something that "subtracts" the 128k and 256k files from the CD rip. The signal level that remains, as a percentage of the CD rip signal level is the total harmonic distortion. I think there's an AU plugin available which will allow you do that in GB, Audacity or Logic, but it's not something I bother with too much.

grorr76
12th January 2009, 06:55 AM
I have a CD, a 128k AAC and an upgraded 256k AAC iTunes Plus file of the same album - is there a way I can tell from these?

yes. open both files up in a waveform editor if both are the same you will see it visually..

stevejay
12th January 2009, 07:02 AM
Edit: I'm a prat.

fompsweeva
12th January 2009, 07:06 AM
yes. open both files up in a waveform editor if both are the same you will see it visually..

The wave form will only show you the amplitude, you need a proper spectral analyzer to see any real info.

markhellewell
12th January 2009, 08:41 AM
My main concern is how the hell did they get all the publishers to re-rip and re-upload their non iTunes Plus content without telling them why?

I imagine that the "special software" they give to publishers uploads to Apple's servers an Apple Lossless file from which iTMS can supply any lower bitrate as required.


Plus I'm listening to some tracks I've upgraded and they sound horrible, worse than the 128k versions.

Perhaps the higher bitrate has exposed weaknesses in the music itself? ;)

mallo
12th January 2009, 09:01 AM
I imagine that the "special software" they give to publishers uploads to Apple's servers an Apple Lossless file from which iTMS can supply any lower bitrate as required.



Perhaps the higher bitrate has exposed weaknesses in the music itself? ;)

I'd agree with this, it would make sence that apple would plan for the future with itunes, and as bandwidth speeds increase, and download limits ect, it would make sense to offer higher and higher quality tracks, so the "special software" most probably allows for this when uploading to apples server.

Vapours
12th January 2009, 09:04 AM
At the risk of sounding naive, I would be very surprised indeed if Apple tried to deceive their customers on this one. It would be one of the dumbest business decisions ever made in the history of IT. They would well know that their customer base includes many serious audio people who would quickly pick up on any such chicanery, and that being caught out would have serious consequences legally, financially, and for their all important reputation. Infinitely easier to just give their customers the genuine 256 bit product in the first place.

marc
12th January 2009, 09:16 AM
You need something that "subtracts" the 128k and 256k files from the CD rip. The signal level that remains, as a percentage of the CD rip signal level is the total harmonic distortion.
Should be possible just by inverting the CD-ripped file, then mixing it with the 128k and then 256k files. I can do this :) The result would be silence for a perfect copy and get louder for something that's more different.

Decryption: Upload some WAVs of a part of the track and I'll get this happening.


At the risk of sounding naive, I would be very surprised indeed if Apple tried to deceive their customers on this one.
Yeah, it doesn't sound very Apple to just upscale, giving worse results. Who knows though, even the best of us screw up sometimes.

decryption
12th January 2009, 11:50 AM
Unshackle your iTunes purchases! | Beau Giles (http://beaugiles.net/blog/2008/12/unshackle-your-itunes-purchases/)

Will just strip out the DRM without re-encoding. :thumbup:

EDIT: Or iMovie '06 will let you import the DRMed files, and then export it as an AIFF file. How stripping the DRM was done ages ago.

EasyTunes just gives an error and I don't have iMovie '06.


The wave form will only show you the amplitude, you need a proper spectral analyzer to see any real info.

Adobe Soundbooth has one, so I can use that?


Perhaps the higher bitrate has exposed weaknesses in the music itself? ;)

The CD version doesn't sound like that.


I'd agree with this, it would make sence that apple would plan for the future with itunes, and as bandwidth speeds increase, and download limits ect, it would make sense to offer higher and higher quality tracks, so the "special software" most probably allows for this when uploading to apples server.

This is not what I've heard from the three people I know that have published music as a label to the iTunes store. It merely uploads the 128k AAC (or 256k AAC if you decide to publish iTunes Plus) you encode on your local machine to Apple's servers.

If I am wrong or have misunderstood, maybe someone can clear it up for me?


At the risk of sounding naive, I would be very surprised indeed if Apple tried to deceive their customers on this one. It would be one of the dumbest business decisions ever made in the history of IT. They would well know that their customer base includes many serious audio people who would quickly pick up on any such chicanery, and that being caught out would have serious consequences legally, financially, and for their all important reputation. Infinitely easier to just give their customers the genuine 256 bit product in the first place.

Exactly! I find it so bizarre Apple would do it, that I want to find out, just to put my mind at ease.


Should be possible just by inverting the CD-ripped file, then mixing it with the 128k and then 256k files. I can do this :) The result would be silence for a perfect copy and get louder for something that's more different.

Decryption: Upload some WAVs of a part of the track and I'll get this happening.


Yeah, it doesn't sound very Apple to just upscale, giving worse results. Who knows though, even the best of us screw up sometimes.

Heading out to buy some iTunes credit so I can upgrade my library at a discount ($130! They don't let you upgrade bit by bit) and to buy a CD or two so I can compare them. Will upload rips soon.

decryption
12th January 2009, 03:59 PM
Here are 15 sec AIFF samples from "Start Her Up (http://heyinternet.com/musicdump/grates.zip)" from The Grates, Science is Golden single and "Strangers (http://heyinternet.com/musicdump/vanshe.zip)" from the album V by Van She.

In the zip file is:
AIFF ripped from the CD using XLD (original store-purchased copy)
AIFF of the DRM 128k AAC (generated by burning the track to an audio CD, then ripping it using XLD)
AIFF of the iTunes Plus 256k AAC (generated using Quicktime's Export to AIFF)

The 15 sec samples were made in Adobe Soundbooth CS4.

NeoRicen
12th January 2009, 04:18 PM
Well I can't tell the difference between any of them, so I'm no use. :(

areal
12th January 2009, 04:52 PM
Here are 15 sec AIFF samples from "Start Her Up (http://heyinternet.com/musicdump/grates.zip)" from The Grates, Science is Golden single and "Strangers (http://heyinternet.com/musicdump/vanshe.zip)" from the album V by Van She.

I could hear a clear difference between the CD rip and the two iTunes files but I couldn't really tell the two iTunes files apart. I did pick the iTunes plus file but there honestly wasn't much difference and I might have just got lucky. I can usually tell the difference between low-quality iTunes AACs and AACs that I rip myself at 256kbps, so maybe there is something bogus going on here. Oh, I renamed them all so I could not know which was which from the title only and I listened on a decent stereo.

Xenex
12th January 2009, 06:31 PM
My main concern is how the hell did they get all the publishers to re-rip and re-upload their non iTunes Plus content without telling them why?

Apple did tell the labels why. That's why the news leaked, and we knew this was coming for months.


At the risk of sounding naive, I would be very surprised indeed if Apple tried to deceive their customers on this one.

If what decryption has noticed turns out to be true, it's not Apple's fault; the labels provide the encoded files. Apple can't be expected to check the audio quality of files on their store, just like old fashioned music stores can't be expected to check the audio quality on every CD.

An aside: I've had more than one occasion where obscure tracks I've downloaded from iTunes have had issues (hissing, pops, static), and one email to iTunes support has provided a fixed download within a few days. So, they don't check everything, but they're happy to help if you raise an issue.

Anyway, anyway, it's going to be interesting to see how this transcoding thing plays out...

melbmac
12th January 2009, 08:03 PM
Will be interested to see the results of this.

I'd like to know what "qualifies" people to comment, are they actually trained in Audio stuff, or do they just consider them selves experts (much like a lot of people that buy DSLR's then claim to be professional photographers)?

Me, I wouldn't comment. Although I do love "quality" sound, I am in no way an expert and I don't have what you'd call "high end" audio gear here (only Yamaha stuff). I can usually tell the difference between mp3's and CD's, but that's not impressive in the least.

decryption
12th January 2009, 08:20 PM
Will be interested to see the results of this.

I'd like to know what "qualifies" people to comment, are they actually trained in Audio stuff, or do they just consider them selves experts (much like a lot of people that buy DSLR's then claim to be professional photographers)?

Me, I wouldn't comment. Although I do love "quality" sound, I am in no way an expert and I don't have what you'd call "high end" audio gear here (only Yamaha stuff). I can usually tell the difference between mp3's and CD's, but that's not impressive in the least.

That's the same reason I'm not claiming there is an actual issue yet, I don't know enough about audio to make that claim (and then get slaughtered if I am wrong).

Someone with more audio skills can be more scientific than me I hope :)

mitty
12th January 2009, 08:41 PM
Reading this thread with interest. I recently did a test... I grabbed out a CD and encoded the songs as 128, 256, and Apple Lossless.

Perhaps it's my 37 year old ears, but I honestly could not tell the difference, at least through my stock standard apple earbuds.

Apple Plus/Lossless/whatever is a rort. Encoding at 256 is more than enough, unless you're Bach or Beethoven or something.

Dubhousing
12th January 2009, 08:47 PM
Apple Plus/Lossless/whatever is a rort. Encoding at 256 is more than enough, unless you're Bach or Beethoven or something.

I certainly don't think Beethoven would care - he was deaf.

yinyang
12th January 2009, 08:49 PM
Apple Plus/Lossless/whatever is a rort. Encoding at 256 is more than enough, unless you're Bach or Beethoven or something.

old Ludwig would have been no good to you in his later years....! :D

edit - beaten to it!!

grorr76
12th January 2009, 08:52 PM
Reading this thread with interest. I recently did a test... I grabbed out a CD and encoded the songs as 128, 256, and Apple Lossless.

Perhaps it's my 37 year old ears, but I honestly could not tell the difference, at least through my stock standard apple earbuds.

Apple Plus/Lossless/whatever is a rort. Encoding at 256 is more than enough, unless you're Bach or Beethoven or something.

you wouldn't be able to tell the difference because the apple ear buds dont produce the frequencies that would allow you to be able to tell the difference between the various encodes you did. A high quality pair of monitors are the key to accessing differences in frequency reproduction as they produce a far greater spectrum that ear buds just cant. there for hearing compression and discarded frequencies becomes far clearer.

mitty
12th January 2009, 08:55 PM
I certainly don't think Beethoven would care - he was deaf.

Yes but he would have been able to hear Apple Lossless ;)

The Drummer
13th January 2009, 08:56 AM
Very interesting.
I just got a chance to have a listen to Decryption's files.

I couldn't distinguish the iTunes versions from each other. They are virtually identical which seems to indicate something might be not be as it seems. As far as I can tell they both feature the same levels of artifacts. They certainly don't sound like different qualities of encoding to me.

It's a bit of a tricky piece to judge everything on because of all the various types of distortion in the original recording but this also makes it a reasonable test of the encoding algorithm.

Rant
13th January 2009, 11:27 AM
This is a very interesting discussion, reminding me of another story. There were reports that the Amazon mp3 of Red Hot Chilli Peppers' Californication album sounded better than the CD !

Apparently it had been taken from better masters than the CD, which is quite possible. Quite a few CDs sound like they were made from LP masters.

What I didn't hear was if anyone compared the Amazon mp3 to the iTunes Californication. AAC should give better reproduction - depending on which master was used, of course. But it's possible that the record companies made better versions for Amazon to gain advantage over iTunes...

Which leads us back to the current discussion - Have the DRM-free versions now on iTunes been compromised in some way? From where did either Apple or record companies get their 256k DRM-free files? And most importantly, how was it done so quickly?

Is it the case that Apple had 256k DRM versions for all these songs and just had to remove the DRM? I don't think so (but I'm not an iTunes user, so I defer to those who are on this point).

Making an AAC (or mp3) is as much an art as any audio process (like mastering) - to do it properly, you'd need to make an AAC master, then tweak the compression to suit the material.

Bulk processing of files will inevitably lead to 'issues'. And I suspect that it will be a matter, as suggested above, for Decryption to take up with iTunes (individually or in batches) to get his library sounding satisfactory (and I know how much trouble that's going to be). Consider it a 'gradual upgrade'.:smash1:

If a few people are going through the process (and I guess that most won't bother), it would be worth a separate thread with multiple people naming offending songs individually, just to get an idea of the scale of the problem and number of users affected.

The more people participating, the more MacTalk has to go to iTunes with & hopefully get higher level management looking into the problem.

As for delivering 'measurable proof', unless there are some compression hackers out there that can read some tags or something to determine the origin of the files, I'm afraid the nature of the compression process is to throw data away - leaving you in the unenviable position of trying to prove a negative.

Anyone out there who can read mp3/mp4 entrails, er... I mean artefacts?

matthewk
13th January 2009, 11:51 AM
Which leads us back to the current discussion - Have the DRM-free versions now on iTunes been compromised in some way? From where did either Apple or record companies get their 256k DRM-free files? And most importantly, how was it done so quickly?

Something like this would not have been done quickly. There would have been months of planning and negotiations, then months of preparation.

I'd presume that Apple has been taking lossless formats from the record companies all along, and then when they were allowed they just did a bulk compression to 256k.

Have a look at TuneCore's requirements for files sent to them: TuneCore: FAQ (http://www.tunecore.com/index/faq#Ripping)

They want WAV for iTunes uploads only. So Apple must demand some form of lossless.

Joe91
13th January 2009, 01:02 PM
Perhaps it's my 37 year old ears, but I honestly could not tell the difference, at least through my stock standard apple earbuds.



That's your problem. The Apple buds hold back any decent quality audio, making it sound horrible.

I've done tests with all the decent IEMs I've had (CK7s, SE210s, SF5s and briefly a pair of SA6s) and the difference is quite clear.

mac_man_luke
13th January 2009, 01:47 PM
Apple Plus/Lossless/whatever is a rort. Encoding at 256 is more than enough, unless you're Bach or Beethoven or something.

I encode everything in lossless where possible

I dont mind 128k ACC / 192k MP3 via earbud from an ipod but if i use quailty headphones or decent stereo then i can tell a compressed file from a mile away - the music looses its life and just turns into a muddle

bullrout
13th January 2009, 05:22 PM
That's your problem. The Apple buds hold back any decent quality audio, making it sound horrible.

I've done tests with all the decent IEMs I've had (CK7s, SE210s, SF5s and briefly a pair of SA6s) and the difference is quite clear.

What are the differences you're hearing?

entropy
13th January 2009, 06:06 PM
This speculation is all very amusing, but just what do you think is going on when you purchase the upgrades to you library? When it says it is downloading, you think it is actually just inflating your current file? Just look at your activity in istat while it is downloading.

I think the problem is that Apple might not be using the best encoders to do the compression, at least for the particular track you are listening to.

decryption
13th January 2009, 06:10 PM
This speculation is all very amusing, but just what do you think is going on when you purchase the upgrades to you library? When it says it is downloading, you think it is actually just inflating your current file? Just look at your activity in istat while it is downloading.

Err no.
I think Apple did a batch re-encode of their current 128k media to 256k and are selling it as upgrades. I haven't really got any conclusive proof beyond a couple of people going "yeah, sounds like it." - hence why I posted it here, for someone that knows more than me to perhaps look into it more?

ford.boy
13th January 2009, 06:15 PM
And entropy: Apple did this with TV shows from a lower res to a higher res at one point.

BoxDog
13th January 2009, 06:16 PM
Yeah, interesting discussion. Has been debated since 2004. It's at least as interesting than the discussion about whether you get a more natural sound if you have your speaker cables pre-heated or gold plated, or the good old vinyl-vs-CD debate. :o)

Thought I'ld give these samples a shot with my in-ear Etymotic ER-4Ps (http://www.etymotic.com/ephp/er4.aspx). Im in no way an audiophile (I'm a bit deaf in one ear from some surgery!!) but got these headphones a couple of years back to get a fantastic sound and still protect my hearing. I figured that by filtering out outside sound and reproducing the full range of frequencies faithfully, it'd stop me wanting to turn up the volume and crank out some horrible treble or bass frequencies.

I could hear a distinct difference between the original CD version and the two AAC encoded files but couldn't really distinguish between 128/256 in both samples. The difference between the CD and encoded versions is pretty much the norm for any sort of mp3 or aac compressed music, in that it sounds "flat". Higher audible frequencies tend to sound "squelchy" and not at all vibrant.

However the differences between even the originals and the compressed samples here were less evident and it judging by the choice of music it was probably because of dynamic range compression (as opposed to compression to reduce data) done in the studio.

In other words Decryption your music tastes are rank!!! :o)

El Guardo
13th January 2009, 06:38 PM
We use to have someone on these here forums from an independent music label - and he commenting on the struggle to get his stuff on the iTunes music store (before it supported independent labels). If anyone would know about the uploading and encoding process...

Or, speaking of knowledgeable people, didn't we have a well known member get an electronica album on the store about six months ago? Again, he/she should know the process up to inclusion on the iTunes store.

Joe91
13th January 2009, 07:01 PM
What are the differences you're hearing?

The most notable difference is obviously the bass. The iPod earbuds (NOT the new in-ears, haven't tried them yet) have pretty much no bass, they sound very tinny and hollow. The IEMs I mentioned earlier actually bring the bass to life, much more punchy and tight - without being unnatural or overwhelming. You can feel the bass in your ears, as opposed to cheap earbuds where you can almost hear the earbud attempt to have bass impact.

Obviously the bass is the most notable difference - but if you listen closer you can quickly pick up much more energetic midrange, and a much more detailed upper end. The listening experience is just transformed from a lump of clay into something beautiful.

I remember when my CK7s carked on me and I had to use the iPod 'buds for a week or two while I gathered the funds to upgrade to a new pair - I honestly couldn't stand the lack of detail and impact in the sound. Once I went from them to my SE210s, and ultimately SF5s - the difference is night and day. I only use my iPod buds for podcasts where I don't need good sounding audio.

To anyone here who has never bought decent audio gear - you are missing out. I have only used midrange IEMs and have been blown away. If you are going to spend hundreds of dollars on an iPod then get some good buds/IEMs aswell. A good place to start would be the Yuin PK1 or PK2.

EDIT: Sorry for thread derailment - but hey, bullrout asked :)

Beau
13th January 2009, 07:18 PM
The application labels and people with the right permissions use iTunes Producer to rip, tag and send your music off to Apple. This is available for pretty much everyone to download directly from Apple if you know the right URL (http://swcdn.apple.com/content/downloads/21/11/061-4583/yzkv3TPNbDBNXJWYWJ7w9xnXRT9fzFPxF8/iTunesProducer.tar). :)


iTunes Producer allows you to prepare your music for sale in the iTunes Music Store. Using iTunes Producer, you can encode your music into AAC format, enter album, song, and artwork information, and send all your prepared files directly to Apple.

Sure you can't actually *do* much with it, unless Apple has pimped your iTunes account into a label account, but it does provide some insight into how the store works. It also shows all the stuff we are missing out on, as the labels are lazy (lyrics, etc).

After importing 'Track 01' from the retail Duosseudo CD (:D) it looks like iTunes Producer *does* rip in Apple Lossless, if MediaInfo is to be believed;
http://img.skitch.com/20090113-t118eremk7ipaqieuachpeirmr.jpg

Beau
13th January 2009, 07:31 PM
And some more bits and pieces;

A music video (Start of Something from Britannia High) in iTunes Plus format (only went plus after the announcement) also has iTunes Plus quality audio, according to MediaInfo. It also reports that it was either ripped and/or sent to Apple in early December, if you believe the 'Date of Original Encoding'.

http://img.skitch.com/20090113-ct2p4ekug134nbajdwe86cjqyw.jpg

And again, 'Track 01' (Together) from the same Duosseudo album, except purchased from iTunes in Plus format, gives a date of August 2008 for date encoded. Again, not sure if this was date ripped or sent to Apple.

http://img.skitch.com/20090113-k6jdrmm3xpasd8u8mw8bye9pdf.jpg

(And no, it was not the date purchased and downloaded with iTunes)
http://img.skitch.com/20090113-xraitfax955355rchhb22f9136.jpg

dotnet
16th January 2009, 02:36 AM
Err no.
I think Apple did a batch re-encode of their current 128k media to 256k and are selling it as upgrades. I haven't really got any conclusive proof beyond a couple of people going "yeah, sounds like it." - hence why I posted it here, for someone that knows more than me to perhaps look into it more?

I think you are right. I've overlaid the 128k and the 256k sample with the inverted CD original, and the resulting leftovers were much alike in amplitude and frequency range:

CD minus 128k AAC:
http://www.dotnet.org/stuff/CD-vs-128k.png

Sonogram:
http://www.dotnet.org/stuff/Sono-CD-vs-128k.png

CD minus 256k AAC:
http://www.dotnet.org/stuff/CD-vs-256k.png

Sonogram:
http://www.dotnet.org/stuff/Sono-CD-vs-256k.png

For comparison I I took one of my lossless tracks and converted it to 128k AAC and 256k AAC. I also up-converted the 128k rip to 256k. The difference between original AAC is much smaller with the genuine 256k rip, and about the same for the 128k and 128k -> 256k rips.

The latter is the case with your samples. You can hardly tell the 128k and 128k -> 256k diffs apart.

Cheers
Steffen.

decryption
16th January 2009, 04:11 AM
For comparison I I took one of my lossless tracks and converted it to 128k AAC and 256k AAC. I also up-converted the 128k rip to 256k. The difference between original AAC is much smaller with the genuine 256k rip, and about the same for the 128k and 128k -> 256k rips.

The latter is the case with your samples. You can hardly tell the 128k and 128k -> 256k diffs apart.

Cheers
Steffen.

Very interesting.

According to Beau up there, the files are uploaded to iTunes in Apple Lossless format. I downloaded iTunes Producer and ripped a CD using it - they are lossless. Apparently this hasn't always been the case (remember, iTunes store started in 2003).

I wonder how I can get enough evidence & proof to write something up about it?

tcn33
16th January 2009, 05:07 AM
The pedantic "audiophiles" on Waffles use Adobe Audition to analyse files as discussed below. Not sure if it would be useful for AAC but just in case:


I've seen a lot of discussion here about how to spot transcodes. Many people have suggested using a spectral analysis from programs like Cool Edit / Adobe Audition / EAC and looking at the 'cut off' point. There is some disagreement about how effective this is, but those who recommend it suggest looking for cut-offs between 1600 Khz as the signature of a 128kbps mp3 source and 2100 Khz as the signature of Lossless.

One counter argument to this 'cut off' level method is that the same cut off which characterises lossy encodes may also be the result of a poor quality recording - a bootleg of a live show or a 'third world' vinyl master.

A number of spectral views have been posted and linked to, but nearly all of these have been analyses of entire tracks... which IMHO is NOT the most effective way to use spectral analysis to detect transcodes.


What I haven't seen anyone discuss is the 'blocky' appearance of the spectral analysis of lossy rips which is noticeable only when you zoom in close enough. IMHO this is a more reliable way to detect whether a file which purports to be lossless has in fact been transcoded from a lossy rip, and may even be a useful way to detect re-encodes from lower to higher bitrate mp3s (although this is much harder whatever method you use).

The image below illustrates what I mean. The track (from an album by Philip Glass) was ripped from CD to flac and a 1 second sample was saved to 320 kbps LAME mp3 and 128 kbps FhG mp3 and then in each case saved again to flac. The spectral analysis was done at full screen on a monitor with resolution of 1280 x 1024. Each of the three strips below is of the same 0.15 of a second.

FLAC / 320 mp3 / 128 mp3 compared
http://gallery.me.com/tnieboer/100092/1/web.gif

And here are bigger strips of the three spectral analyses. The zoom level is the same - bigger simply means that what is shown here is around 0.5 of a second - and NOT the whole track!

FLAC
http://gallery.me.com/tnieboer/100092/2/web.gif

320Kbps LAME mp3
http://gallery.me.com/tnieboer/100092/3/web.gif

128Kbps FhG mp3
http://gallery.me.com/tnieboer/100092/4/web.gif