HydrogenAudio

Hello, I have just made this account. Long time user of foobar, occasional lurker here on the forums. I have been trying to get to the bottom of this problem I've had with finding a decent audiochecker program. I hope this is the right sub-forum to post this in. I just typed out my inquiry on another site, so I'll just copy and paste that here:

Can somebody help me out, or direct me to a site I should go to to ask about the best audiochecker program? I have lossless audio files on my computer that I want to verify their authenticity with. It's mostly stuff I bought off of bandcamp, 7digital, qobuz, and sites like that depending on the artist. I typically use a spectogram (spek is a great program) which is a great way to snuff out the obvious transcodes. But sometimes the spectogram isn't enough. I've noticed that some 320kbps mp3's or even iTunes AAC's (256 kbps) still have the high 22 khz ceiling, so they're virtually identical under a spectogram. I also have Adobe Audition if this helps.

I have always used a program called Audiochecker by Dester. This is a good program but it hasn't been updated since 2006, and for some reason I can't get it to write a .log file for me when it's done. I guess I could just copy and paste the results, but when you go into the "show complete log" option it doesn't include the overall CDDA percentage it shows in the main menu. It's not a huge inconvenience, just annoying.

Then I started using a program called Lossless Audio Checker, which is one of the first things that shows up when you google for audiocheckers. It writes .log files for you. Great, right? Not really. I've noticed that no matter what I put in, it always comes out saying "Clean". Even files that come up as MPEG on the other program.

I also tried a program from y-soft.org (defunct site) called aucdTectManager. This one is really good, I think. It seems to be out of comission as far as updates though. However, I can't figure out how to get it to write a .log file! It can only do one file at a time and it analyzes the md5 checksum and stuff like that. I just want one convenient .log for CDDA checking! If anybody has used this program I'd like to know how to make it write a .log for a series of files.

Then just before typing this, I tried a program called tau analyzer. Unfortunately it only reads CD's, so that's not going to work (unless I burn my files to a CD first which I don't feel like doing).

Zoom in on the spectrogram more to see it thinning out. Also look at the difference between stereo channels. Make a SoX command-line that trims first and last 10 seconds or so of audio, where the level is lower. The spectrogram will be built much faster than if the whole file was processed. Zoom in to at least 25-40 samples per second (-X). Run the command on a set of files as if transcoding them, then look at the generated images. I use this command for stereo files (replace "input.wav" with a variable that your frontend provides).

sox.exe input.wav -n trim 0 10 -10 remix 1 2 1v0.5,2v-0.5 spectrogram -X 48 -y 257 -Z -10 -z 120 -t "input.wav / L,R-Side / first 10 sec, last 10 sec" -w Hann -o "input.wav.png-48-lrs.png"

It's a huge topic with many misinformation in the web. Audio checking and analyzing is not a simple task. And many informations are mixing up CD analysis, Release analysis (e.g. web shop remasters) and audio file analysis (upscaled fake releases). There is a reason for this unbelievable and highly expensive hardware tools TV companies use to test if a third party production company delivers upscaled audio and video material to reach their minimum deliver requirement conditions.

So it depends: Are you looking for all of them or only integrity check  of CD ripped files? Or do you want to find upscaled fake Audio in your library? Etc, etc. And it becomes even harder when the difference between the awaited result and the true result is small enough, that it becomes almost "similar" in detection. And the tricks used for upscaled audio or CD integrity faking files became even "better" over the years. It will be hard to get a detecting analysis correctly done on a smartly noise shaped 320 kps mp3 upscaled to a 44.1/16bit FLAC file with a faked CD integrity, especailly when it tries to simulate an early CD from the 80s where the spectrum of the masters was almost on the same of well shaped 320 kps mp3.

I typically use a spectogram (spek is a great program) which is a great way to snuff out the obvious transcodes. But sometimes the spectogram isn't enough. I've noticed that some 320kbps mp3's or even iTunes AAC's (256 kbps) still have the high 22 khz ceiling, so they're virtually identical under a spectogram. I also have Adobe Audition if this helps.

Exactly. Spectrum isn't enough. Even if sox and Spek are awesome tools and can be easely added to a workflow, for example via run services in foobar. We use a similar line of code like j7n's for quick testing (EDIT: but we use ffmpeg here since sox seems dead and last version was from 2015 with a lot of security and overflow issues reported). But in case when the original was a "bad" and cheaply produced master of an indie band from an island it will be almost impossible to find artefacts and differences clearly enough between the original CD track and a smartly compressed lossy high bitrate copy. And when it comes to digitalized Vinyl (and the mistakes they all do) it even gets a much bigger and messed up story ...

To be honest I haven't found yet a tool which 100% successfully created postive false results from my extreme testing on them with my audio test files. The most accurate test results you can get is on the CD level: comparing checksums and if you have the original data you can compare them to. I know, a quite antiquated Audio resolution ...

Here is a good explanation which saves me writing even more here about how complex the issue can become to correctly interprete spectrogram data to determine upscaling. Read the best voted detailed answer over there: https://superuser.com/questions/842575/can-ffmpeg-detect-the-actual-bitrate-of-an-audio-file

And here another link to a German site (use a translate plugin of your browser to read in English) testing a non-free but promising audiofile checker application XIVERO MUSICSCOPE and demonstrating the complexitiy of tests needed to really determine suspicious files: https://www.mobilefidelity-magazin.de/macht-den-hi-res-schwindel-sichtbar-xivero-musicscope

While I am not 100% sure if this tool can find really ALL suspicious files with minimum errors in it and while I personally do not like Non Open Source applications where I can not look under the hood, but I am sure that this tool works better than any of those you have tested. And I think it is worth its 30 bugs.

EDIT: oh no ... after writing this, I had to find out that they stopped their development ... https://www.xivero.com ... but it says the application is still downloadable ... ? Hm. But I think it isn't free now, just the downloads for the former registered users. But I tested and installed the demo. The first 30 seconds audio file data are free in the test version without license. So maybe it is worth a look. Just to let know ...

When evaluating quality we don't usually care to determine the exact bitrate of the original, but to detect a transform codec stage at all. So the problem is simpler. It is usually possible to tell if the bitrate was low, high or if another process, such as noise reduction has been applied by observing how the bandwidth changes in response to the signal level. The test tones were steady unlike music.

It is unsurprising that in the linked example there were frequencies present all over the the spectrum even at low bitrate because the steady tones were unusually loud, and thus selected by the encoder for keeping, relatively simple to encode, and generated clipping both before and after encoding (0.8 =-2 dB, and 4 were summed). Regular music or sound effects would reach high levels at high frequencies only occasionally, and only those instants would be encoded. The spectrograms are not zoomed in enough, but even here I can see gaps in the 64 kbit/s example.

Yes it is possible to tell if the bitrate was low or noise reduction introduced observing how the bandwidth changes in response to the signal level on 0815 upscales. Buts thats a rather perfect test and ideal detection scenario. In reality those cases wouldn't even fail with the simplest tools. But the other ones which are not easely to detect they would fail. In reality it means: there is no drag drop "test the bunch of files in the folder and forget about it". This is what the OT was referring to when saying that the tests he did not always work and he asks if there's a better tool for it. And that's what I am referring to when I say the topic is more complex than just checking the Spectrogram and Bandwidth in some cases. And I am an audio mastering engineer.

I am pretty sure most simple tools won't succeed (detecting) when I give a "musically" upscaled file from 320 kbs mp3 to 441/16 FLAC, where the upscaler knows what he is doing. Musically upscaling means to introduce elements which blend the artefacts realistically (Dynamic Expander, Multiband Frequence Dynamic EQ, just to count some of them). Similar to upscaling and repainting in photoshop from a LoRes jpg. That's what I mean when I say that the best check would be a comparision with the original. Which is mostly not available in these cases.

To the frequencies all over the spectrum at the linked test comment: I just gave that link to generally demonstrate that it is not that simple like some websites mistakenly try to say (not more) with their simple go-to: look where it cuts off and you will know what kind of mp3 it was. It is not that simple. While I agree with the loudness aspect on high frequencies in the test over there, I would strongly recomment not to create the readers assumption that this is a rare case and by accident of the wrong test only. The hardly to control fuzziness of the so called additional "Harmonics" play an important part here and also on music while encoding and decoding. In any frequency scale and level. And change on every device or tool used for it. Look at those sweep tests https://src.infinitewave.ca/ - And they play an important part by EQing and Compressing (I talk about musically compressing dynamics, not the codec).

We should not forget about the basic reason why music listeners are interested in such tests and tools: To be sure that the audio file has not been corrupted in a lossy way. And for me there is no such tool at the moment, which would successfully find every error with 100% success rate. Only the obvious ones with a success rate of maybe 70%.

I use MusicScope or RightMark Audio Analyzer (http://audio.rightmark.org/products/rmaa.shtml)

@soundping -> yeah "MusicScope" looks promising but have you heard of that they closed the doors? Can you still run a legit copy and do you know if there is an opportunity to still purchase a late license? I think this 30 bugs are well invested for at least another 2 years compatibility with codecs and file resolutions etc.

RMAA is rather a signal flow test suite. It's a great tool indeed, but from what I know not for testing files isn't it?

I like very much this one: auCDtect Task Manager

Very reliable.

I like very much this one: auCDtect Task Manager

auCDtect was developed by y-soft.org (site down) and has been discontinued. Please remove the link you posted since we can not guarantee that the software installer provided there is harmless. The site do not provide an original hash for this zip file offered there. Here an alternative.to link with similar apps and the original link to their dead website: https://alternativeto.net/software/aucdtect-task-manager/

I wanted to participate just to help. It won't happen anymore.
Thank you for your kindness.

p.s.: IT'S IMPOSSIBLE DO EDIT MY PREVIOUS POST. PLEASE, SOME ADMIN, DELETE MY PREVIOUS POST AND MY PROFILE TOO. Or at least ban my profile so that I don't accidentally re-enter here.

Thanks in advance.

I wanted to participate just to help. It won't happen anymore. Thank you for your kindness.
p.s.: IT'S IMPOSSIBLE DO EDIT MY PREVIOUS POST. PLEASE, SOME ADMIN, DELETE MY PREVIOUS POST AND MY PROFILE TOO. Or at least ban my profile so that I don't accidentally re-enter here. Thanks in advance.

Oh no. I am very sorry. I do not wanted you to feel like that. If I used wrong phrasing please consider it caused by that English is not my native tongue. I just wanted to inform that such links should be handled with care. Sorry if you felt bad about it.  Please accept my apolgize, That was not my intension. I just wanted to be careful about external links that  can harm computers.

perhaps we should approach this from the opposite view

first, find the best upscaling or re-mastering program for music, then see how they go about producing the lost portion to make it better again?

one of the equivalent in the visual world be waifu2x, which works for both video and static picture

These restoration tools extend the bandwidth by copying over parts of the lower spectrum or adding harmonic distortion (exciter), and filling in some remaing gaps with noise. On an analyzer you can see that features such as blocks from the encoding or the interference from a TV monitor get doubled. It would take a high level of added noise to cover them up.

The gap filling from Stereo Tool's delossifier module didn't sound like an improvement to me. There are processes that output continuous spectrum (ac3, dirac pitch shifter), but still sound bad at extreme settings. Some time ago I listened to the output of Zynaptiq Unchirp. It added an in impressive emphasis on transient sharpness. Its smoothing of the blockiness of "musical noise" stabilized background sounds, but also made them noticeably duller and happily removed sqeaks from guitar that were part of the original sample but similar in character to artifacts.

Waifu2x looks impressive on non-anime images. It restored cat's eyes and whiskers.

These restoration tools extend the bandwidth by copying over parts of the lower spectrum or adding harmonic distortion (exciter), and filling in some remaing gaps with noise. On an analyzer you can see that features such as blocks from the encoding or the interference from a TV monitor get doubled. It would take a high level of added noise to cover them up.

The gap filling from Stereo Tool's delossifier module didn't sound like an improvement to me. There are processes that output continuous spectrum (ac3, dirac pitch shifter), but still sound bad at extreme settings. Some time ago I listened to the output of Zynaptiq Unchirp. It added an in impressive emphasis on transient sharpness. Its smoothing of the blockiness of "musical noise" stabilized background sounds, but also made them noticeably duller and happily removed sqeaks from guitar that were part of the original sample but similar in character to artifacts.

Waifu2x looks impressive on non-anime images. It restored cat's eyes and whiskers.

one of the amazing capability about Waifu2x is its ability to upscale video and picture. ex DVD to BD, shitty scanned to better looking scan, or BD to 4K/UHD
though not always giving best result, but it's probably the top program out there for these type of things despite so many new programs, AI assisted applications and algorithms had became available since Waifu2x came out long ago

so I thought perhaps audio world would have some equivalent programs like this

I have always used a program called Audiochecker by Dester. This is a good program but it hasn't been updated since 2006, and for some reason I can't get it to write a .log file for me when it's done. I guess I could just copy and paste the results, but when you go into the "show complete log" option it doesn't include the overall CDDA percentage it shows in the main menu. It's not a huge inconvenience, just annoying.

I just registered to post a comment here. :) It is heartwarming to see that you are still using my software, as I am the developer of Audiochecker. The reason behind the development stop is that I had less and less time to maintain it, also I switched to .NET from Delphi and I decided to release the future developments. I want to emphasize that my software is just a shell to be able to manage the files and directories without the hassle of writing BAT files or similar, as the software uses the auCDtect engine to determine the result. Why the log file cannot be written, I have no idea. By the time I wrote the software, Windows XP was used, so it might be something with incompatibility.
Since then the world changed a lot and also multicore systems became common. I had an idea to rewrite the whole thing using modern methods and also take advantage of parallel processing, but I have very little spare time. However, I'm a bit surprised to see that after 15 years the software is still in use. :)

I am also still a regular user of 'audiochecker' these days. I see it as an extra 'checkup tool' which still can have some value in combination with other tools. Of course I know as a 'standalone' tool it has its limitations.

Not very long ago I was able to obtain new, really lossless files of an album I bought in FLAC-format. The first FLAC-files I received were almost sure from a lossy source, and I was able to strenghten my case because of analysis with several tools, among which audiochecker.

Therefore I would welcome an updated version of the tool, suitable for modern systems.

Btw: not being able to write the logfile is indeed a known issue over here on recent Win10 configurations. On Linux via Wine it works fine.

Small update: when using the MS Windows compatibility function Audochecker is still able to save the logfile. Just set it on MS Windows XP or MS Windows 7.

Check out Loseless Audio Checker. They claim it is backed by  "research papers".

https://losslessaudiochecker.com/

Check out Loseless Audio Checker. They claim it is backed by  "research papers".

https://losslessaudiochecker.com/

I know this tool as well, it's functionality is limited compared to Audiochecker. I also remember having some technical issues with it on both MS Windows and Linux, can't remember the details. As this tool hasn't received an update in the last 5 years I don't think it is worth testing it again to fresh up my memory however.

Check out Loseless Audio Checker. They claim it is backed by  "research papers".

https://losslessaudiochecker.com/

This software is useless.
https://www.audiosciencereview.com/forum/index.php?threads/flac-bitrate-resolution-quick-way-to-check.23458/post-790377

This software is useless.

It is.
The concept is nice but the implementation suffers.
In practice it looks at the highest frequency.
If this is below <20 kHz it thinks it is  mp3.
Anyway, the good news is that all this audio "checking" software is like us.
The moment the bit rate is high they are unable to discriminate between lossless and lossy.
Just like us  :)

Exactly.  Placebo effect.

61kbps aac detected as "clean" :))
https://hydrogenaud.io/index.php?topic=111736.msg922129#msg922129

Are there really any known algorithmic giveaways available to give a good conclusion, except comparing to the encoders' known lowpasses?

(One should also be careful about selected counterexamples, as you cannot really surely tell that it isn't part of what the artist deliberately sent off. But we knew that, eh?)

M4A file, Bitrate 61 kbps. Cut-off at 20.5 kHz.

Writing application : fhgaac v03.02.16;VBR=2

How are you setting up the cutoff frequency ? With Winamp ? Because I can't find such an option in any fhgaacenc version.

How are you setting up the cutoff frequency ? With Winamp ? Because I can't find such an option in any fhgaacenc version.

I used the foobar front end with auto settings. When the preset quality is low enough it will switch to SBR mode and render beautiful spectrograms, provided the source audio files have enough high frequency content.

(One should also be careful about selected counterexamples, as you cannot really surely tell that it isn't part of what the artist deliberately sent off. But we knew that, eh?)

Yeah, for example some "noise reduction" plugins in DAWs, when used with sources like noisy analog tapes. Depends on user settings and how the plugins are designed, it is possible to create gradual lowpass with spectral discontinuity, not only they can fool checking software, the resulting files can even look similar to some lossy codecs in spectrograms.

Some vintage synthesizers like the Roland SC/JV series used 32kHz synth engines and they will also result in obvious lowpass in spectrograms.

I used the foobar front end with auto settings. When the preset quality is low enough it will switch to SBR mode and render beautiful spectrograms, provided the source audio files have enough high frequency content.

Thanks for the information, I'll do some testing :)

Since I'm using foobar portable, do you know by any chance where I should copy the fhgaacenc folder ? Or should I just drop all of the needed files in the Encoders folder ?

I am using portable version too. Just copy the highlighted files to the encoder folder.

Yep, I've figured it out, the reason why I've removed this question from my post.
Thanks for the answer, it might help someone else one day ;)

I've been using spek recently: https://github.com/alexkay/spek

Quote from: Porcus on 2021-05-29 13:30:45

(One should also be careful about selected counterexamples, as you cannot really surely tell that it isn't part of what the artist deliberately sent off. But we knew that, eh?)

Yeah, for example some "noise reduction" plugins in DAWs, when used with sources like noisy analog tapes. Depends on user settings and how the plugins are designed, it is possible to create gradual lowpass with spectral discontinuity, not only they can fool checking software, the resulting files can even look similar to some lossy codecs in spectrograms.

Some vintage synthesizers like the Roland SC/JV series used 32kHz synth engines and they will also result in obvious lowpass in spectrograms.

Like this:

Hi everyone, Does Using foobar2000, and is 'warning pannel' which tell you 'corrupted file etc..' is enough to sort 'corrupted folders, and' clean folders'?

My matter is, recently my mind was away, so my external hard drive, which have sometimes been disconenected because of usb cable.

By copying mass audio  folders it goes sometimes in 'errors'.

Usually I have good strategy and solves issues like this, 'manually' and the strategy includes a final result that's makes safe and sur that dedicate folders are 'safe' I mean no corrupted files in these.
But today i'm a bit lost.

Hi everyone, Does Using foobar2000, and is 'warning pannel' which tell you 'corrupted file etc..' is enough to sort 'corrupted folders, and' clean folders'?

Not quite sure what you want. If you use foo_verifier, the "verify integrity" function will give you a list of errors and warnings.

Good thing:
* You can click the status or warnings column header, get all corrupted on top, mark them (using shift+arrow), right-click, move (or copy!) them with the move entire folder content option to, say, "D:\suspicious\%path%", recreating full path on a different drive.

Not so good thing:
* Warnings may be nothing special, or they may be serious - and especially for the mp3 format, serious issues may be hidden in a haystack of minor issues. That goes if reported and decoded lengths are slightly off - which was common with certain software. I had such a thing when a mobo dropped the drive mid-writing, and mp3 files just got different audio. Tough thing to find those that were really corrupted.

Possibly even worse:
* If the file is so corrupted that fb2k cannot even make sense of it, you won't even have it loaded into the playlist. That is, you can have a folder with one corrupted file, but where fb2k won't recognize it as corrupted because it won't even recognize it as an audio file.
But, with a recent version that shows errors in library scanning ... you might be able to pick up more.

Quote from: kwam202 on 2021-09-16 10:48:15

Hi everyone, Does Using foobar2000, and is 'warning pannel' which tell you 'corrupted file etc..' is enough to sort 'corrupted folders, and' clean folders'?

Not quite sure what you want. If you use foo_verifier, the "verify integrity" function will give you a list of errors and warnings.

Good thing:
* You can click the status or warnings column header, get all corrupted on top, mark them (using shift+arrow), right-click, move (or copy!) them with the move entire folder content option to, say, "D:\suspicious\%path%", recreating full path on a different drive.

Not so good thing:
* Warnings may be nothing special, or they may be serious - and especially for the mp3 format, serious issues may be hidden in a haystack of minor issues. That goes if reported and decoded lengths are slightly off - which was common with certain software. I had such a thing when a mobo dropped the drive mid-writing, and mp3 files just got different audio. Tough thing to find those that were really corrupted.

Possibly even worse:
* If the file is so corrupted that fb2k cannot even make sense of it, you won't even have it loaded into the playlist. That is, you can have a folder with one corrupted file, but where fb2k won't recognize it as corrupted because it won't even recognize it as an audio file.
But, with a recent version that shows errors in library scanning ... you might be able to pick up more.

Yes thanks.
I wrote a long text here and was deleted.

I' ll look.
But I imagine very well the global thing.

That's why I do manually everything.

But there it's the mess.

My mind was away. Makes me worry.

A lot of 'corrupted files'.

I feel like what time. I will need to sort everything.

The only thing ?

Most of tracks in my foobar2k playlist with "?" are corrupted.
Then there is half of the track playing.


I'm really in fear of an infinite sorting folders.
Plus some folders on hdd 'said' safe's are also there.

The mess.

Don't look further, wi'm a bit in the sound, and computer.

There are so many ways in sound, and 'digital'.. .. Difficult to find a solid software, too mucb different uses for individuals ? Even if a lot of parameters are useless.

The mess.

Well, what do you have? Backup?

To be honest I haven't found yet a tool which 100% successfully created postive false results from my extreme testing on them with my audio test files. The most accurate test results you can get is on the CD level: comparing checksums and if you have the original data you can compare them to. I know, a quite antiquated Audio resolution ...

It's not an antiquated solution if it's literally the only solution. I get the appeal of pirating and the thrill of getting free stuff, but CD's are the only way to truly own music.

I mean not to go full vault boy here, but do people honestly think that in an economic collapse or any other sh*t hit the fan scenario companies are going to maintain these giant servers with your copy of Stairway to Heaven somewhere in the cloud?

Why do people who wouldn't trust Google, Amazon, & Apple with their cookie data and browsing data, trust them with all of their digital purchases?

By copying mass audio  folders it goes sometimes in 'errors'.

It's hard to answer your question properly with such little technical detail.
However I would definitely recommend a free program called "Teracopy" rather than just using Window's explorer's inbuilt data transfer tool.

Teracopy has a range of options for what to do in case of an emergency, including settings on what to do when you are not present at the PC and write errors occur.

Teracopy also verifies all the files after writing, so you can be sure that you have a bit-perfect copy.

hi! so um, i found this a while back and its browser based, but i dont think it works anymore :(, but regardless, you can still try :)
https://finalizer.com/analyzer?a=m31705&b=117&compare=true&mview=triple&ref=a

If you dont want to look at graphs Audio Checker is good.
I tested LAC and Audio checker with best quality upscaled MP3.
I tricked LAC but Audio Checker caught me.
Audio Checker is old and as far as I know there no further development but it works
I have tried over and over different files and I can't trick it.
Also for linux users it works in wine if you do it right.
My 2 cents.

I've been using spek recently: https://github.com/alexkay/spek

Me too. I use it as run service context item in foobar, so I can check a doubtful file quickly on right mouse click. Additionally I use the (sometimes slow but awesome sox based) spectrogram from spider monkey panel in a second tab and compare both results. SoX is still besides ffmpeg a daily basis tool here on cmd. Apart from that I have some serious other tools (prof. audio mastering analyzis tools) here to check audio regarding data loss if it is very important to be sure.

A lil bit offtopic: The problem is, as you can read in this thread, people have interest in development of undetectable upscaling (sadly), which can cause problems in the future to detect upscaled files, if they succeed in these efforts. Which is problemaitc in times of internet song sales and to trace back from what source the material has been created. And which does not create any win on the real quality of the compressed material. To be honest I see no (positive) reasons in upscaling. As an audio and video engineer I just can say what I sad multiple times: there is no such thing, no such magic like bringing back the original lossless file. The term "compression" is misleading. It is technically a measurable cut off from data which CANNOT be brought back. They only can get simulated and this is a fraud in my opinion because it hasn't anything to do with the original audio or video material. It is no upscaling, it is rather a simulation closer to animation than to real material. This is why we frame old low-scale video when we have to embed it in high-scale documentaries. This is always the better choise vs upscaling. No matter how good the algorythm is.

I came to the conclusion that's impossible to recognize upscaled or modified files if you are not buy at least Digital or better CD.
Even with CD you have to trust Reseller and EAC with secure + AccurateRip.

There are a lot of Modification bad people can do, flac rencode to mp3 and back is easy recognizable.
It would be easy to take FLAC files from WEB Shop modifie (some reverb here, some eq tweaks) and upload
fake EAC Log to AccurateRip.
You fake it multiple times with different CD Drives and you get high confidence level.

So even EAC Secure + AccurateRip is problematic when it comes to recognizing the original and you don't own the CD.

You have to own the CD and rip it yourself and thats usually how far you get when it comes to perfection.

Even an Original CD Rip from you can have a low confidence level.
That mean nobody else upload this media and/or it's repress on factory cause simple checksum change but audio will identical.

The only way to avoid all the stress is to buy or rent the CD.

I personally buy on Bandcamp CD+Digital or from trusted reseller CD only.

I had such a great program around 10 years ago. I can't remember what it was called. It was great at confirming MP3's that were converted to FLAC. Is there any current solution? So many dodgy FLAC's out there. With everything going to streaming crap, it's getting harder and harder to keep a lossless collection :)

aucdtect, as mentioned earlier in this thread?

I had such a great program around 10 years ago. I can't remember what it was called. It was great at confirming MP3's that were converted to FLAC. Is there any current solution? So many dodgy FLAC's out there. With everything going to streaming crap, it's getting harder and harder to keep a lossless collection :)

There's one thing to keep in mind: a large percentage of modern sold CDs/FLACs are remasters with a horribly low dynamic range and muffled instruments and old "shitty" MP3s often sound a ton ... better.

Don't chase numbers, pursue audio quality.

Another issue is that with more DIY around, the artist might have temporarily converted the file to MP3, not realizing (or not bothering to care about) that it is lossy.
Then some digital audio vendor demands WAVE or FLAC upload, so ... OK, WAVE they get.

aucdtect

I'm afraid all these type of programs are like us. Struggling to detect lossy from lossless.
https://www.thewelltemperedcomputer.com/SW/AudioTools/Detect.htm

Not sure this qualifies as best or even good, but this old proggy is quite nice for a number of things. It has flaws and I'm not sure how well it will run on current Windows (using it on W7 here, and it already has some flaws).
Hope you find it useful.
https://web.archive.org/web/20131005073901/http://www.neillcorlett.com/informer/

Faking the Funk is bad, like really bad.

It erroneously detected over 10% of my FLAC tracks as MP3's ranging from 320 down to ... 96Kbps. That's insane.

Here's a 96Kbps "MP3" file.