Hi friends!

I am very strict with the sound quality. When I record a CD I try to do it from .wav files, so the quality keeps equals. But sometimes, when a person gives me a homemade CD, I feel curious about the quality, then, I rip some tracks to .wav files and I open them in Audition. I do a Frequency Analysis (Alt+Z). If the CD was burned from a lossy format (e.g. .mp3, .ogg or .wma files), frequencies beyond 18 kHz are attenuated.

But that proof is not always infallible. Recently, I did a lossy compression with WavPack around 260kbps. I drop the file over Audition and I found the frequencies beyond 18kHz are amplified! and I am sure, there was loss of quality but the graphic in Audition seems contrary.

Is there a real proof to know if a CD track was recorded from a losless source or not? It is extremely important for me.
Thanks a lot in advance.

Pd.: I beg your pardon for my English.

The only other method I can imagine is to rip a few tracks using a quality based encoder like the newest version of LAME. I have noticed that if I rip from a CD that was burned from lossy sources LAME will hang around at the lowest bitrate allowed (say 192 kbps). Of course that seems to be the case only for very low quality sources.

On the other hand you could do what I've done: adopt the very pragmatic view that if it SOUNDS fine it is, in fact, fine. No sense in getting yourself all worked up over something you can't hear under any circumstances.

There's no 100% reliable way to tell whether a CD was burned from a lossless source unless you have access to the original CD (or another lossless source).

Tools like tau analyzer might provide pretty good guesses, but they won't be 100% accurate.

Also, I have some CDs that have a frequency cutoff at ~18KHz (those are CD versions of LPs).

Tau Analyzer is a great app for doing this. It may guess, but it's damn good at it. http://www.true-audio.com/Tau_Analyzer_-_C...ticity_Detector

I advise otherwise; this does not seem as though it would be a good or very accurate way to do what the original poster desires.

More concerned with detecting lossy sources/stages in live recordings (which is more difficult) but the forum in which this is a sticky is a great place to learn.

I advise you to read the whole post where I qualified the idea.

And?

Surely added quantisation noise and other artefacts of lossy coding could actually increase the audio's complexity in the eyes of following encoders. But I'll try this on one of my commercial CDs that I know for a fact was mastered from MP3, due to the lowpass, just to entertain you! Still, I don't feel that this is a very optimal solution, or one to recommend when there are better alternatives available.

I'm not sure why my suggestion bothers you so much. You seem bent on ignoring the fact THAT I SAID IT'S PROBABLY NOT RELIABLE in my first post.

The audio is not made "more complex" by the presence of artifacts. I have re-ripped a number of discs that had had MP3 sources and all have exhibited the behavior I described earlier. If the audio has already undergone data compression then all the low-level data is gone.

I was also unaware of the Tau Analyzer before this thread. If I really wanted to be sure that the audio was uncompressed I'd just buy an original copy of the album.

Whenever I backup a CD or find an old backup I've burned I also wonder (and am concerned) about the quality and whether it was created as an exact copy of the CD or if it was created from MP3 or some other lossy compression.

The basic, and I'll admit, less than fool proof, way I usually attempt to do this is to see if freedb or whatever info service Windows Media Player uses can find the info of the CD.

Apologies forehand if this is a newb question, but something I've never thought about: Can you create an image file of a CD using MP3 or other lossy formats? I never used to keep these when backing up a CD. I just rip it straight to MP3. So, if I need to create a new CD from my backup I usually end up burning it from the MP3 files. Then that CD's info cannot be found when a freedb (or otherwise) lookup is done.

However, I have had CDs that are so rare and cherished that as soon as I purchase the CD I make an exact copy of it. So, when I find these old CDs that is the method I use to determine if it's created from a lossy compression or if it's the real deal.

Jed

freeedb recognition cannot serve as indication whether the CD is sourced from lossily compressed material or not. All that is necessary for the CD to be recognized is to reproduce the original's layout - tracks durations and the order of them.

Without ever educating myself on the data structure of a CD I've always figured that a CD has some sort of unique ID and this is what a system like FreeDB uses to look up the info (artist, title, track names, etc) for a CD. It seems like a DB search based on track duration and order could be a time consuming process and easily return multiple results.

OK then, try ripping an original CD to MP3, then burn those MP3s to a disc. Then pop it back into your computer and have freedb fetch the details. You'll get the same results as you did with the original disc.

Close, but not exactly. I just tried this with Wilco's "Yankee Hotel Foxtrot". The process I used:

1) Used EAC to rip the CD to wave files
2) Then used LAME recommended encoding settings to encode as MP3
3) Removed the original wave files
4) Used LAME --decode to decode MP3 files to wave files
5) Used Burrrn to burn an audio CD with the tracks in the proper order

Now, I did this on a PC not connected to the internet. The reason for this is that it is possible that a ripping program would look up the info of the CD and cache it. Then after encoding and decoding and bringing the waves back into the program to write them to an audio CD it would recognize them and associate the cached info with those files. It would then do the same when the CD was burned and forever associate that info with the CD, incorrectly making you think it was actually giving info based on just what was on that CD.

I've noticed that both EAC and Windows Media Player work similarly to this. You'll notice that once EAC looks up the info for a CD it will never go out to the internet and ping the FreeDB database again for that info - it caches it.

So, to get back to my process. I then went to a PC that was connected to the internet, put the newly created CD in the drive and started up EAC. EAC brought up no info on the CD. When I went to "Database" -> "Get CD Information From" -> "Remote FreeDB" it took about ten to fifteen seconds and then returned to me five different choices of Wilco's "Yankee Hotel Foxtrot" to choose from. Which I must admit was not bad - I was impressed how quickly it found the info and how it didn't happen to return any other CDs that it possibly could be. However, it seems the probability is fairly high that duplicates incorrect choices could be returned.

I'm not sure how many CDs are in FreeDB, but a number anywhere from a few hundred thousand to a couple million would not surprise me. With that amount I'm sure there are some with identical number of tracks and matching track durations.

But to get back to the experiment. When going to yet a different PC connected to the internet that has never seen (as to cache the info of) the Wilco CD and placing it in the drive and then starting up EAC, EAC immediately (a couple of seconds) pinged FreeDB and pulled back all the correct info (artist, title, tracks, etc). I didn't have to go to "Database" -> "Get CD Info", etc... from the pulldown menus and it didn't give me a choice of five different choices of the Wilco CD info.

Also, similarly, when placing the newly burned disc in Windows Media Player it didn't automatically find the info. However, when placing the authentic Wilco CD in the drive Media Player found all the correct CD info in a couple of seconds.

That was an interesting test. EAC gave better results than I had expected. I wonder how it would do with a lesser quality burning program that automatically puts 2 seconds between each track...

Why so elaborate? Test it by:
- rip whole CD as .cue/.wav
- encode the WAV to MP3
- decode MP3 to WAV
- burn the .cue/.wav

100% match between Gracenote/FreeDB/etc everytime.

Yes, if you have the cue sheet you should be good to go. My (too lazy to go and check it out for sure) theory is that the cue sheet includes some sort of unique ID, or close to unique ID, that FreeDB and other services use when looking up the CD. If someone wishes to correct me on this issue I'm all ears.

I didn't want to burn using a cue sheet because the vast majority of people I know who have MP3 collections don't care about cue sheets. They swap around CDs with albums on them that are just straight MP3s. When they go to burn an album they just burn all the MP3s. The filenames seem to usually always start with a two digit track number so it's easy to keep all the tracks in the correct order...