Spectrograms before and after conversion to flac

[Ankin]

Hey, everyone.

I have an .aac file extracted from an .mp4 video using My MP4Box GUI. Supposedly, no changes were introduced during demuxing. Here's the spectrogram:

Now, my mp3 player cannot play .aac, but it does play flac. So I convert this audio into flac using Audacity 2.0.1 (Level: 5; Bit depth: 16).
Here's what I get, when I open the newly created .flac file:

It is my understanding, that a .flac file should contain the exact replica of the original waveform. So what are those artificats at 15-22k throughout the length of the audio?

thanks

[bat999]

Hi
What's the spectrogram look like if you use FFmpeg/WinFF instead of MyMP4BoxGUI and Audacity?

ffmpeg -i filename.mp4 filename.flac

[Slipster]

FLAC itself is lossless, yes. Any artifacts you see are almost certainly the result of Audacity's conversion from AAC to a bitstream before the re-encode to FLAC.

You're also wasting an enormous amount of space by converting the AAC file to FLAC, as well as the fact that you're gaining nothing from doing so apart from compatibility with your playback device.

I'd highly recommend converting to MP3 instead, but please don't use Audacity to do this as it makes a complete balls-up of MP3 conversion by forcing you to use CBR mode which is by far the least efficient method in terms of quality versus bitrate/file size.

If it were me, I'd install Foobar2000 and use the LAME MP3 encoder with a VBR setting of -V2. It's incredibly unlikely that you'd hear any difference between this and the original AAC file in a double-blind listening test, plus you'd save yourself an enormous amount of storage space.

It's also worth largely ignoring spectrograms in favour of actually listening to stuff. We hear with our ears, not with our eyes.

[poisondeathray]

dithering setting in audacity

https://forum.videohelp.com/threads/328465-Why-does-Audacity-always-save-audio-files-differently

[Slipster]

Well caught that man! I was forgetting that this still hadn't been fixed by defaulting to 'none', as well as MP3 encoding still defaulting to CBR despite the LAME codec developers specifically tuning for and recommending VBR over CBR for more than a decade.

[poisondeathray]

Why not convert directly to mp3 as Slipster suggested ?

Many ways to do this - lamexp, foobar2k, megui, ... dozens others

[Ankin]

dithering setting in audacity
https://forum.videohelp.com/threads/328465-Why-does-Audacity-always-save-audio-files-differently

Edit -> Preferences -> Quality -> High Quality Conversion Dithering - NONE
Now the spectrogram of the saved .flac file is identical to that of the sourse .aac file. Thanks a lot!

[jman98]

Well caught that man! I was forgetting that this still hadn't been fixed by defaulting to 'none', as well as MP3 encoding still defaulting to CBR despite the LAME codec developers specifically tuning for and recommending VBR over CBR for more than a decade.

Look, I've got something LAME developers can suck on. Since the highest possible bit rate in MP3 is 320 Kbps, assuming that non-standard higher bit rates are avoided, it is absolutely freakin' IMPOSSIBLE for any VBR MP3 to surpass 320 CBR MP3 in terms of quality. Now if you want to debate that on storage challenged playback devices that saving a few piddly MB via VBR might be helpful in terms of getting more music on the device or that on a well encoded VBR file at a high enough bit rate that most listeners can't tell the difference between it and CBR, you may have a point. But VBR MP3 is not problem free on some playback devices and with disk storage at 3 TB and rising, the space savings requirements that made VBR get invented in the first place for MP3 have disappeared.

In my mythical 320 CBR MP3 file example, since any VBR encoding will have some parts of it that use a lower bit rate than 320 and they cannot go above 320 since that is the maximum, you are now arguing that lower bit rates are better than higher ones, a rather bizarre position to take based on logic alone. I think LAME developers are arguing a point that may be lost on you - namely that few can tell any difference between VBR and CBR if the bit rate is high enough so you might as well save the tiny amount of space that VBR gives you because it MIGHT (who knows?) be important on SOME playback devices that are space challenged. But it's not better in terms of quality and that certainly cannot be the reason they suggest it. Please spare me the apples to oranges comparison of comparing something like VBR files with an average bit rate of 256 Kbps which can have sections that go as high as 320 Kpbs vs. something like a 256 CBR file or a 198 CBR file. If LAME devs want to argue that 256 VBR is better than 256 CBR, yes, they are probably right. But that goes out the window if you use 320 CBR. VBR cannot exceed that value at any point and will have sections that are under it.

[Ankin]

Hi. What's the spectrogram look like if you use FFmpeg/WinFF instead of MyMP4BoxGUI and Audacity?

Never used them, might give it a try sometime.

Why not convert directly to mp3 as Slipster suggested ?

You're also wasting an enormous amount of space by converting the AAC file to FLAC..

Yes, I am aware of the space requirements of the .flac format. The thing is the source audio is already crappy, I would hate to introduce any further distortion by transcoding it into another lossy format. Anyway, thanks for the suggestion.

[Slipster]

Look, I've got something LAME developers can suck on. Since the highest possible bit rate in MP3 is 320 Kbps, assuming that non-standard higher bit rates are avoided, it is absolutely freakin' IMPOSSIBLE for any VBR MP3 to surpass 320 CBR MP3 in terms of quality.

False. The encoder uses different psychoacoustic models in CBR and VBR mode, the latter having been heavily tuned over the past decade whilst the former has remained largely stagnant, so it's entirely possible for a high bitrate VBR encoding to trounce a 320kbps CBR encoding under some circumstances, and they often do.

Now if you want to debate that on storage challenged playback devices that saving a few piddly MB via VBR might be helpful in terms of getting more music on the device or that on a well encoded VBR file at a high enough bit rate that most listeners can't tell the difference between it and CBR, you may have a point.

Yes, the point being that VBR can match CBR320 for perceived quality in most double-blind test cases at -V2 which averages around 60% of the overall file size.

But VBR MP3 is not problem free on some playback devices and with disk storage at 3 TB and rising, the space savings requirements that made VBR get invented in the first place for MP3 have disappeared.

The very few playback devices that won't play back a VBR file correctly aren't standards-compliant as it's been a part of the relevant spec for over 15 years. I realise that that's no consolation to anyone unlucky enough to own such a device, but it is, strictly speaking, broken. I fail to see how the '3TB and rising' argument is relevant when the entire purpose of a high quality lossy encode is to sound transparent to the end-user. If this can be achieved in 60% of the space (which it can), where's the sense in spending the same amount of time encoding it and making it 40% bigger for no audible gain?

In my mythical 320 CBR MP3 file example, since any VBR encoding will have some parts of it that use a lower bit rate than 320 and they cannot go above 320 since that is the maximum, you are now arguing that lower bit rates are better than higher ones, a rather bizarre position to take based on logic alone. I think LAME developers are arguing a point that may be lost on you - namely that few can tell any difference between VBR and CBR if the bit rate is high enough so you might as well save the tiny amount of space that VBR gives you because it MIGHT (who knows?) be important on SOME playback devices that are space challenged. But it's not better in terms of quality and that certainly cannot be the reason they suggest it.

If you think that saving up to 40% in terms of file size and using a more mature and finely tuned psychoacoustic model that is capable of providing better transparency in general is irrelevant to the whole argument of CBR versus VBR, then yes.

Please spare me the apples to oranges comparison of comparing something like VBR files with an average bit rate of 256 Kbps which can have sections that go as high as 320 Kpbs vs. something like a 256 CBR file or a 198 CBR file. If LAME devs want to argue that 256 VBR is better than 256 CBR, yes, they are probably right. But that goes out the window if you use 320 CBR. VBR cannot exceed that value at any point and will have sections that are under it.

Which is also a largely irrelevant argument as you're assuming that the encoder always needs more bitrate than it's going to use when in VBR mode. Once you've exceeded the bitrate required to achieve the target quality for any given frame of audio, the rest is wasted in CBR mode, and that goes against everything that lossy encoding stands for, ie, space saving.

If you want to stick to a deprecated encoding mode that's no longer developed with any rigour and strongly recommended against by the codec developers, feel free, but I'll certainly not be recommending it except for live streaming work where a fixed and defined bitrate is often the preferred method of encoding.

[Ankin]

The whole argument comes down to this statement:

The encoder uses different psychoacoustic models in CBR and VBR mode, the latter having been heavily tuned over the past decade whilst the former has remained largely stagnant, so it's entirely possible for a high bitrate VBR encoding to trounce a 320kbps CBR encoding under some circumstances, and they often do.

If it's true, then you're right and jman98's wrong. If it's false, then you're wrong and jman98's right

[Slipster]

The whole argument comes down to this statement:... If it's true, then you're right and jman98's wrong. If it's false, then you're wrong and jman98's right

It's definitely true, and you're welcome to go and talk it over with the current LAME developer on the Hydrogenaudio forums if you like. I'd like to make it clear that I'm not interested in any, "I'm right and he's wrong!" arguments. I'd just like the truth to be known, and jman98's not the first person I've come across who isn't familiar with LAME's internal workings. Let's be honest, not many people besides the developers themselves and very keen followers are, so let's not be having a downer on the guy, m'kay?

[Ankin]

Now that I'm aware of dithering, it kinda poses another question. Wikipedia says that dithering is used when digital audio is reduced to 16 bits for pressing onto a CD..
When I produce a flac file in Audacity, the output bit depth is 16 bits. Now that dithering is off, does this mean that I should go to Edit -> Preferences -> Quality -> Sampling and set the Default Sampling Format from 32-bit float to 16 bits as well?

[aedipuss]

The whole argument comes down to this statement:... If it's true, then you're right and jman98's wrong. If it's false, then you're wrong and jman98's right

It's definitely true, and you're welcome to go and talk it over with the current LAME developer on the Hydrogenaudio forums if you like. I'd like to make it clear that I'm not interested in any, "I'm right and he's wrong!" arguments. I'd just like the truth to be known, and jman98's not the first person I've come across who isn't familiar with LAME's internal workings. Let's be honest, not many people besides the developers themselves and very keen followers are, so let's not be having a downer on the guy, m'kay?

personally i think the difference between vbr and cbr is mute, as they both suck. compared to listening to live music or a vinyl record on decent audio equipment, cds are awful, and any mp3 made from a cd sounds like mud.

[Slipster]

personally i think the difference between vbr and cbr is mute, as they both suck. compared to listening to live music or a vinyl record on decent audio equipment, cds are awful, and any mp3 made from a cd sounds like mud.

Very funny!

[pandy]

personally i think the difference between vbr and cbr is mute, as they both suck. compared to listening to live music or a vinyl record on decent audio equipment, cds are awful, and any mp3 made from a cd sounds like mud.

Very funny!

yep, especially when You know that vinyl is mastered from same digital source as CD - i enjoy such claims when pickup is MM type and vinyl is played more than 10 - 20 times.

[aedipuss]

personally i think the difference between vbr and cbr is mute, as they both suck. compared to listening to live music or a vinyl record on decent audio equipment, cds are awful, and any mp3 made from a cd sounds like mud.

Very funny!

yep, especially when You know that vinyl is mastered from same digital source as CD - i enjoy such claims when pickup is MM type and vinyl is played more than 10 - 20 times.

wouldn't know anything about current vinyl, haven't bought any since ~ 1980. thorens tuntable, grace wooden tonearm, b&o mmc20cl cart., macintosh tube preamp, marantz tube amp.

[hello_hello]

Look, I've got something LAME developers can suck on. Since the highest possible bit rate in MP3 is 320 Kbps, assuming that non-standard higher bit rates are avoided, it is absolutely freakin' IMPOSSIBLE for any VBR MP3 to surpass 320 CBR MP3 in terms of quality. Now if you want to debate that on storage challenged playback devices that saving a few piddly MB via VBR might be helpful in terms of getting more music on the device or that on a well encoded VBR file at a high enough bit rate that most listeners can't tell the difference between it and CBR, you may have a point. But VBR MP3 is not problem free on some playback devices and with disk storage at 3 TB and rising, the space savings requirements that made VBR get invented in the first place for MP3 have disappeared.

Except..... when encoding the bitrate isn't the only factor determining quality. It's admittedly the most important one but not the only one. Hands up anyone who's specified a quality other than the default while encoding a constant bitrate MP3? How many people even know you can? Yet oddly enough many people seem to know how a 128k MP3 will sound without needing too fuss over anything as mundane as actually listening to it (I'm not having a good at you, just thinking out loud).
I'm not arguing 320 CBR encoding would theoretically keep more of the original audio goodness than a lesser bitrate but it's no doubt in a way similar to video encoding. There's a point where increasing the bitrate produces fairly diminishing returns in perceived quality, and there must be some point a mere mortal can longer see differences after the bitrate has been increased.
-q n Bitrate is of course the main influence on quality. The higher the bitrate, the higher the quality. But for a given bitrate, we have a choice of algorithms to determine the best scalefactors and huffman encoding (noise shaping). -q 0: use slowest & best possible version of all algorithms. -q 2: recommended. Same as -h. -q 0 and -q 1 are slow and may not produce significantly higher quality. -q 5: default value. Good speed, reasonable quality -q 7: same as -f. Very fast, ok quality. (psycho acoustics are used for pre-echo & M/S, but no noise shaping is done. -q 9: disables almost all algorithms including psy-model. poor quality.

I don't know if the default being -q 5 is wrong or out of date but I seem to get -q 3 unless I specify a quality. Most (video) encoder GUIs seem to use -q2. The little bit of testing I did a while back seemed to indicate -q3 encodes at about twice the speed of -q2 so I assume the encoder uses the extra time wisely?

Even over at hydrogenaudio, where no doubt many of the regulars have ears so golden they can reliably predict which company supplies your electricity and where and how its generated simply by listening to it's effect on audio.... even the golden eared experts there seem to have reached a consensus on MP3 bitrates and the effect on quality.

http://wiki.hydrogenaudio.org/index.php?title=LAME#Recommended_encoder_settings
"-b 320 is an alternative to the VBR settings above. This CBR mode will maximize the MP3's bitrate and overall file size. The extra space may allow for some parts of the audio to be compressed with fewer sacrifices, but to date, no one has produced ABX test results demonstrating that perceived quality is ever better than the highest VBR profiles described above."

[hello_hello]

personally i think the difference between vbr and cbr is mute, as they both suck. compared to listening to live music or a vinyl record on decent audio equipment, cds are awful, and any mp3 made from a cd sounds like mud.

I find myself needing to disagree, and therefore ask, was your "any mp3 made from CD sounds like mud" conclusion reached through extensive ABX testing or is it primarily due to a placebo effect? Once again, even the folks over at hydrogenaudio seem to think MP3 can achieve transparency. What did they miss while performing their ABX testing which hasn't eluded you? They even describe the -V4 (~165 kbps) preset as "being close to perceptual transparency". Even "close to perceptual transparency" sounds a fair way above a "mud" rating to me.

Not that I have the best audio equipment in the world, or the most transparent headphones, but I did do a bit of ABX testing myself at one stage. CD to wave file vs 192kb VBR MP3. Could I reliably play one and tell you whether it was the CD or the MP3. Not a chance. Could I reliably pick which was which when comparing them back to back..... well if I really paid attention I'm pretty sure I could sometimes hear very minor differences between the two, but not a "this bad MP3, this good CD" kind of difference. Just a very minor difference. Certainly no sound of mud.

Poor old MP3. Play a CD and if it sounds bad you'll no doubt be offered a variety of explanations except for the possibility it's a crappy sounding CD, play an MP3 and it sounds bad..... even when plugging an MP3 player into a sound system while continuing to use the same EQ which makes those ear buds sound bearable.... MP3 is apparently the problem there too.

[sanlyn]

personally i think the difference between vbr and cbr is mute, as they both suck. compared to listening to live music or a vinyl record on decent audio equipment, cds are awful, and any mp3 made from a cd sounds like mud.

Very funny!

I don't know that it's funny, but it's generally true.

yep, especially when You know that vinyl is mastered from same digital source as CD - i enjoy such claims when pickup is MM type and vinyl is played more than 10 - 20 times.

I don't believe that's true. Many recording companies still master on analog tape. If you can't hear the difference, it won't matter. Matters to me. Almost all digital audio sucks in comparison. My external DAC off my CD player makes a little improvement, but not that much. If you've been playing your vinyls on that Technics turntable and its aluminum arm, destruction began on the first play. Forget about 20.

AR -> modified Rega -> Grado -> Denon -> Dynaco -> B&K (or sometimes PSB or Beyerdynamic). Sure sounds better to me. Enjoy.

[aedipuss]

not really the thread for a debate over analog vs. digital music. but i allow that for the vast majority of people mp3 is "good enough". most think that the fm radio in their car is high fidelity also.

I find myself needing to disagree, and therefore ask, was your "any mp3 made from CD sounds like mud" conclusion reached through extensive ABX testing or is it primarily due to a placebo effect? Once again, even the folks over at hydrogenaudio seem to think MP3 can achieve transparency. What did they miss while performing their ABX testing which hasn't eluded you? They even describe the -V4 (~165 kbps) preset as "being close to perceptual transparency". Even "close to perceptual transparency" sounds a fair way above a "mud" rating to me.

ok so mud is a hyperbole, but unless everything in your audio pipeline is class "a" it's quite possible to not hear an appreciable difference. even the noise transistors make switching on and off in most inexpensive consumer amps can be a factor.

[pandy]

yep, especially when You know that vinyl is mastered from same digital source as CD - i enjoy such claims when pickup is MM type and vinyl is played more than 10 - 20 times.

I don't believe that's true. Many recording companies still master on analog tape. If you can't hear the difference, it won't matter. Matters to me. Almost all digital audio sucks in comparison. My external DAC off my CD player makes a little improvement, but not that much. If you've been playing your vinyls on that Technics turntable and its aluminum arm, destruction began on the first play. Forget about 20.

AR -> modified Rega -> Grado -> Adcom -> Dynaco -> B&K (or sometimes PSB or Beyerdynamic). Sure sounds better to me. Enjoy.

Nope, analog tape is niche - even small audiophile recording companies use digital recording techniques - usually 96 ksps 24 bit which is kind of industry standard - simply this is OK if You know what to do with audio - if this is not enough then You can use higher sampling rate - even DSD (SACD) recordings are made in that way - which is quite funny when You consider war between DVD Audio and SACD - SACD processing imply use PCM even for simple mixing which can't be done in digital domain as for PCM. Key to good digital audio is correct level (case 0dBFS), correct dithering and correct noiseshaping.

Destruction in vinyl recordings is unavoidable (unless You use contact-less way to play such as laser turntables or modern high resolution optical readers) - 20 times played with very high end characteristic stylus tip will degrade quality badly - deterioration of the groove is unavoidable. So how frequently You can listen Your favorite album, how many of copies You need to have?

Don't understand me wrong - i like for example CRT's however i can't agree that "digital audio sucks" - if it sucks this is not because of technology but because of trends in music industry (infamous loudness wars etc). I can understand that some people like distortions of the vacuum tubes amps/preamps ("warm feeling"), i can understand that some people like vinyl's - that's OK - some people like it some not i respect both points of view however digital technology of reproduction offer much higher quality (almost transparent) and what is more important it can even emulate by DSP all distortions created by tubes and/or vinyl to imitate those techniques of reproduction.

With correct DSP You can fit in CD, audio with levels bellow -130÷150 dBFS (on 16 bit;44.1ksps) - imagine what You can do with 192ksps 24 bits.

And once again - some people don't like transparent audio - they found audio reproduced in that way as "clinically cold" however this is purely subjective way of what You like - not objective classification.

[Ankin]

Did You play Your vinyl with liquid lubricant (like isopropylic alcohol) to reduce wearing of groove and lower overall noise floor? Or perhaps You use one of the nice laser turntables?

Do you always type 'you' with a capital Y?

[Slipster]

personally i think the difference between vbr and cbr is mute, as they both suck. compared to listening to live music or a vinyl record on decent audio equipment, cds are awful, and any mp3 made from a cd sounds like mud.

Very funny!

I don't know that it's funny, but it's generally true.

So the ~99% of listeners who hear no difference whatsoever between a LAME MP3 encoding at -V2 and the lossless original in double-blind listening tests all have defective equipment and/or hearing then? Interesting.

If you can take an ABX test and come up with some reliable results then I'll start taking your claims seriously, but the only cases where I've seen ABX test results showing that a difference could be reliably heard were from young testers (<20 years old) who openly admitted to having clinically diagnosed abnormal hearing of one kind or another.

even the noise transistors make switching on and off in most inexpensive consumer amps can be a factor.

The only consumer amplifiers where there are any transistors in the audio chain "switching on and off" are Class D designs. What percentage of typical consumer amps are Class D? I'd gladly take a guess at less than 1% with the other 99% being Class B or Class A/B. If you're talking about crossover distortion in simpler Class B designs then it may be audible in a very poorly designed amplifier, but designs that bad are very few and far between nowadays. 30 years ago, maybe. Now? Only if the amp fell out of a Christmas cracker.

[hello_hello]

Fair enough..... admittedly when I read comments in forums which exaggerate the loss of quality through converting to MP3 it does tend to annoy me somewhat. I remember being involved in a debate in another forum after one poster stated he could tell just by listening to it whether the CD player in his car was playing an original CD or one which contained a bunch of MP3s. I mean really.....

I have no doubt my audio gear isn't of a high enough quality to let me hear any appreciable difference caused by converting to MP3, so I wouldn't offer an opinion there. I do find myself wondering how many people who've offered an "MP3 is crap" post at some stage would own gear of sufficient quality too, or whether many of them in fact own gear which is messing with the audio while decoding it and they assume it's the format's fault.

What about other formats such as AAC or even AC3? The ones which don't seem to be inflicted with a "crap" label. Relative to MP3, are they any better/worse when it comes to transparency? Not owning gear of a quality which allows me to listen for myself, I'd assume encoding using any lossy format would have the potential to cause a perceivable difference now and then.

Which of course leads me to an inevitable question regarding the point at which perceivable difference = bad difference. Given there's so many things in a playback chain which can effect the sound waves which eventually reach our ears.... I suppose an example might be comparing one brand of "super-expensive, this is as good as it gets" speakers with another. They'll probably both sound great, you might even prefer one, but they're not going to sound the same, ever. Okay so we all know speakers colour the sound, and in a perfect world lossy audio formats wouldn't, but it still illustrates my question.

[pandy]

What about other formats such as AAC or even AC3? The ones which don't seem to be inflicted with a "crap" label. Relative to MP3, are they any better/worse when it comes to transparency? Not owning gear of a quality which allows me to listen for myself, I'd assume encoding using any lossy format would have the potential to cause a perceivable difference now and then.

Similar to MP3 - they are lossy way to compress audio - some can be better tuned some not. Most important is what You hear and what You like - if from Your personal subjective point of view is OK then it is OK.

And You can go higher with MP3 bitrate - lame offer:

Constant Bit Rate (CBR)
-b n set bitrate (8, 16, 24, ..., 320)
--freeformat produce a free format bitstream. User must also specify
a bitrate with -b, between 8 and 640 kbps.

or perhaps use a lossy codec optimized from audiophile point of view (like .MPC - Musepack)

Do you always type 'you' with a capital Y?

To show that i respect my interlocutor.

[FulciLives]

The people who argue MP3 can be or very nearly is transparent are delusional.

Now just read this with a critical mind:

You are going to take a 16-bit 44.1kHz PCM WAV file (standard CD audio) and encode it in such a way that actual parts of the audible signal are being tossed out according to some voodoo mumbo jumbo psychological method based on how the brain works (and as I understand it the physicality of the human ear is 'adjusted for' as well) and all this is being done in such a way that it is 100% or very nearly transparent?

Bullshit.

- John "FulciLives" Coleman

[pandy]

The people who argue MP3 can be or very nearly is transparent are delusional.

Nope - it can be quasi transparent for 99% of human population and this is fact that count - majority see (hear) no difference then model for lossy compression is OK, and due of limitations (remember fact that MP3 is a technology from 80/90 - design with respect to available processing power in those times).

However today we can push technology higher and offer with same bitrate higher quality or with lower bitrate comparable quality thanks to more complex algorithms.

That's all - at some point anyway we achieve saturation - perceived sound can't be distinguish from original - then we have quasi-transparent lossy compression.


And what is more important - how many of You guys knows how sounds normal instruments - ie how many of You hear for example cello in various acoustics, how many of You can distinguish between various cellos, made for example from various wood types or by different cello master craftsman's.

This is very good starting point when we talking about audio.

Please accept that our ears and brains hearing artificial sound from very early beginning - younger generation probably don't know what is natural sound for many musical instrument - all they know is how those instrument sounds after processing and lossy compression and thei brains adapt to this sound... and their brains can find not compressed natural sound as something weird and unnatural...

[Slipster]

The people who argue MP3 can be or very nearly is transparent are delusional... Bullshit.

So you too have carried out your own independent ABX tests to come to this conclusion? I'd be very interested to see your log files please. You must be one of the 'special' ~1% who can tell a difference under controlled testing conditions.

[Slipster]

What about other formats such as AAC or even AC3? The ones which don't seem to be inflicted with a "crap" label. Relative to MP3, are they any better/worse when it comes to transparency? Not owning gear of a quality which allows me to listen for myself, I'd assume encoding using any lossy format would have the potential to cause a perceivable difference now and then.

AAC, AC-3 and MP3 are all crap if used incorrectly, ie, forced into a non-default encoding mode or starved of bitrate. They're also all capable of perceptual transparency (ie, sounding the same as the lossless source) under most circumstances if used correctly by a competent individual at the encoding end.

That's where it all falls down I'm afraid as the encoders easily find their way into the hands of those who don't understand how to use them, or think they do but actually don't. It took a concerted campaign on the part of another forum to get Amazon to stop deliberately botching their AAC encodings a couple of years back. Similar pressure has had to be applied to Apple in the past to get them to stop making 'broken' AAC encodings too, so it's not just private individuals who get it badly wrong.

There really is no excuse for it on the part of the professionals apart from laziness and/or ignorance when the codec developers are often perfectly happy to discuss these matters if people make a little effort and contact them. The alternative, particularly for private individuals, is to visit a forum with strict rules regarding any claims made and evidence of any such claims having to be provided, as that's the only way to get fact-based advice.

If the individuals here who claim that lossy audio encoded with the intent of being perceptually transparent (in particular, LAME MP3 at -V2) is crap actually came up with some fact-based evidence to support their claims then we could help them to find out why it sounds crap to them, although I have my suspicions that none of them have actually carried out any controlled tests with encodings made with the developer's recommended settings so have nothing of any real value to offer.

If these individuals would like to provide the results of their ABX tests showing that they can hear a difference under controlled conditions then I'm happy to apologise for accusing them of misleading others, although this will still mean that they're one of the 1-in-100 'golden ears' individuals who genuinely can hear a difference when extensive testing has already proven multiple times that the rest can't regardless of how good their audio kit is.

PS Just for fun, I've uploaded two 44.1kHz 16-bit WAV files HERE. They're 'zipped', so you may need to install WinRAR to extract them. They may or may not be the same, so some of you may or may not hear a difference between them. I'd be interested to know whether any of you think you can and, if so, in what way you think they sound different.