Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Not a dev., but I got an idea for a component. (Read 6355 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Not a dev., but I got an idea for a component.

Hey there !

I was wondering if anyone else though that it would be a good idea.

In the playlist, right-click on the file --> check with auCDtect algorythm.

Now I know that auCDtect is not 100 % precise and you have to be carefull with the results, but sometimes it is spot on.

Anyway, your thoughs ?

Thanks

Alan


Not a dev., but I got an idea for a component.

Reply #2
If you didn't pirate music you wouldn't need such a component.
It's only audiophile if it's inconvenient.

Not a dev., but I got an idea for a component.

Reply #3
You're wrong Kohlrabi. Webshops sell transcodes as well. I bought several transcodes at Beatport, Juno and Trackitdown.

Not a dev., but I got an idea for a component.

Reply #4
You're wrong Kohlrabi. Webshops sell transcodes as well. I bought several transcodes at Beatport, Juno and Trackitdown.
How do you know that they didn't just receive another master, which they sold to you? Not being CD sourced is neither a necessary nor a sufficient criterion to determine if something is a transcode. In the end you just have to trust the webshop you buy your music from.
It's only audiophile if it's inconvenient.

Not a dev., but I got an idea for a component.

Reply #5
Well, when I bought wave with file that I have on mixed CD I can directly compare spectrals  . Unfortunately sometimes spectrals don't match - CD has full frequency and wav from Beatport has flat cut off at 18,5 kHz... Do I still have to convince you? What if waves have very specific distortions, that let's you even guess what encoder was used?

Not a dev., but I got an idea for a component.

Reply #6
^are you using the tool as a basis to get refunds then? do these stores accept it?

i'm just curious. i can't think of any other practical use.

Not a dev., but I got an idea for a component.

Reply #7
Do I still have to convince you? What if waves have very specific distortions, that let's you even guess what encoder was used?
It's not about convincing me, it's about convincing yourself, and not drawing wrong conclusions from insufficient data. You will have a hard time convincing me that spectrals or lowpasses are a sure-fire way to identify lossy encodes, lowpasses are at most necessary. My point is that all you can do is trust the store, like you have to trust the mastering engineer/label when buying a CD.
It's only audiophile if it's inconvenient.

 

Not a dev., but I got an idea for a component.

Reply #8
@ Kohlrabi - if you have just 2 copies of the same track (of course one is just excerpt from mix CD, other is full lenght) and one of them has permanently visible lowpass applied - would you still believe webshop? Other thing is that after my complaints certain track was resupplied by label. Original copy looks like it was encoded to 128 kbps mp3 and then decoded to wave, and resupplied copy has perfect spectrals. Sorry, this is genuine evidence.
@ marc2003 - yes. If I decide to buy mp3 then I accept the fact that quality will be compromised - I paid less so I got lower quality. But if I decided to pay more for perfect quality then any quality compromise is unacceptable. No matter if I hear this difference or not. I get what I paid for or I get refund. And I always got refunds when I got evidence that reported track was fake lossless. Fortunately it happened only about 10 times and I bought music on-line more than 150 times. In nearly every faulty purchase there was just single track that was faulty, one time there were 3 tracks.

The problem is that Juno removes faulty tracks and restore them only if they got proper copy, but Beatport just give refunds and still sell faulty tracks and NEVER removes nor resupplies them. I even found tracks that were just mislabeled. Juno removed them, Beatport simply did nothing. Be careful for what you pay for. We have beautiful days, when everything is so simple to obtain from the Internet. You don't have to go from Poland to Berlin or Amsterdam to buy good music on vinyls (like it was 10 years ago). Now you can buy everything on the Internet and have it just in a while. But products that you can buy not always have proper quality.

Not a dev., but I got an idea for a component.

Reply #9
@ Kohlrabi - if you have just 2 copies of the same track (of course one is just excerpt from mix CD, other is full lenght) and one of them has permanently visible lowpass applied - would you still believe webshop? Other thing is that after my complaints certain track was resupplied by label. Original copy looks like it was encoded to 128 kbps mp3 and then decoded to wave, and resupplied copy has perfect spectrals. Sorry, this is genuine evidence.
What makes you certain the lowpass wasn't in the master track, and the extended spectra on the CD are just due to CD mastering?

I'm not saying that you're wrong, you're most probably correct in pointing out the faulty tracks, I'm just saying that a lowpass is not conclusive. Without the master track you cannot know which of the releases is a genuine lossless track.
It's only audiophile if it's inconvenient.

Not a dev., but I got an idea for a component.

Reply #10
There is simple way to distinct mastering lowpass and encoding (transcoding) lowpass. Filters used in lossy encoding are very steep polyphase IIR filters (transition band is very narrow). Lowpass on spectrogram looks like bandwidth was just cut with knife at certain frequency. Values of this "frequency knife" are also very specific - for example 16,5 kHz (128 kbps), 18,5 kHz (192 kbps), 20,5 kHz (320 kbps). I even tested it some time ago with LAME 3.97 with --verbose parameter. Cut off frequencies were displayed in LAME window, additionally they are written into LAME Tag and can be checked in mp3 files with apropiate software (like EncSpot) or with foobar's spectrogram (for both mp3's and decoded waves).
Low pass filters used in mastering are mostly gradual FIR filters - you can see that energy gradually lowers with increasing frequency. If you look on spectrogram of such file you can see that peaks which comes from hi-hats are diffusing with increasing frequency. In opposition - hi-hats of transcoded files are just cut with "frequency knife" at very specific frequencies.

Of course this isn't true for 100% of files. Sometimes (very rarely) mastering is done using steep IIR filter and sound is cut off above certain frequenciy. But in such case cut off frequency doesn't correspond with any typical cut off freq known from mp3 encoding. Additionally you must keep in mind that lowpass isn't the only transcoding artifact. If you look on spectrogram and see that sound becomes more granulated in function of increasing frequency, you should know that this is also typical lossy encoding artifact - connected with ATH (adaptive threshold of hearing) technique and intensive removing quieter sounds/lower amplitudes above certain frequency. Sometimes you can even see 2 horizontal, sharp lines on spectrogram. Above first there is still sound, but it is very granulated/chunky in function of increasing frequency (image is less saturated, you can see much more background color). Above second there is no sound. This is typical for mp3 encoding - accordingly to my experience this occurs mostly with FhG encoder.
Other question is that everything is quite easy to find when lossy encoding was done using mp3 encoder. HQ AAC encoding (like the one used on iTunes) is nearly impossible to detect.

Not a dev., but I got an idea for a component.

Reply #11
There is simple way to distinct mastering lowpass and encoding (transcoding) lowpass. Filters used in lossy encoding are very steep polyphase IIR filters (transition band is very narrow). ...


Things can be much more complicated than you imply.

There are a number of tools out there of varying quality and masterers with varying tastes, not every mastering decision is made based on purely technical merits.  Assessing the providence of a release is at times impossible even for industry professionals.

Also there are  often a number of legitimate remastered releases out there, some of which were, say, analog, then DSD, then CD...  Who knows what choices were made at each stage for each different release.

With a remix/remaster from the original elements you may well have a different spectral envelope, etc.

On the Deep Purple "Machine Head" SACD release they used the original elements for the new surround mix, but since so many people were very familiar with the album, they chose an early stereo master which is clearly of inferior quality from a technical standpoint (e.g. tons of rolloff), but is still the intent of the masterers.

There is a lot of misinformation out there from amateur audio archeologists.

I agree 100% with Kohlrabi, you have to trust your source.

Not a dev., but I got an idea for a component.

Reply #12
Trust your source... Yes, especially when after my complaints they provided proper copy, which clearly implies that initial one was just a reencode. What is funny - while Juno has proper copy now, Beatport still offers that reencode.
Other problem showing that there is serious problem with quality of tracks sold in webshops are crackles/glitches (sounding like those which occur when you rip very scratched CD - NOT like those caused by excessive use of dynamics compressors and limiters), dropouts and the best - selling excerpts from mixes as full tracks. That was my very first purchase at Trackitdown... Fortunately that happened only once, but crackles and dropouts occur even more often than reencodes. And they aren't problems connected with transmission errors - I download EVERY file at least twice and perform MD5 crosschecking between those copies. So it is your choice whether to trust your source or being careful about quality.
You say too many theoretical things about mastering decisions and different tastes. Distortions introduced by mp3 encoding are very specific - I described them above. While I was studying I learned a lot about perceptual coding. We even compared decoded streams from Philips DCC, Sony ATRAC and 192kbps MP3 sources using Audio Precision System Two (APx 525) on labs.
In the end please think about it - what is the probability, that some mastering engineer will use very sharp filter with cut off frequency corresponding exactly to the one that is standard for some mp3 encoding preset? Why he used cut off exactly at 16,5 kHz (128kbps)? Why he did additional "mastering techniques" so the spectrum shows typical mpeg distortions? Why he mastered in this way the file that was intended to be distributed digitally, but he made "normal" mastering for file that was used on mix CD? Yes, those were really strange mastering decisions, his taste is very "varying one"... The answer is that someone made a mistake at some stage of distribution process. I don't know who and on which stage but I know that I don't want to pay for someone else's mistakes. I never expect refunds - I always ask for replacing faulty copies and letting me download it again when they will be available. But this is impossible at Beatport and not always possible at Juno (but it's still better than never). So mostly it ends up on fcuk1ng refund and having fcuk@d up copy of your favorite track. So what I need is to buy nice turntable, soundcard with proper quality analog inputs and serach for vinyls at Discogs... I already started buying some vinyls, turntable is bit more expensive thing so maybe next year... Yeah and soundcard...
Again - it is your choice whether to "trust your source" or being careful about quality. Fake lossless, glitches, dropouts, incomplete tracks - this is what may happen when you use webshops. In case of glitches and dropouts you should be just careful while listening to your music, in case of fake lossless you can help yourself with spectrograms, programs like aucdtect and detailed knowledge. For me foobar, SoX and aucdtect are very powerful tools to analyze such problems.