[Whitigir]

Wait, Noise shaping and Noise filtering, Anti Aliasing are not considered DSP ? Those are the majority of Digital Filterings

Yes, there are pulsation errors as these codes are read and calculated with the timing by the algorithms and the Digital signal processing techniques. A values given would have to be rounded up or rounded down, if it exceed the values of the resistors to determine the Pulses. That is why R2R and NOS are the least accurate ways to process digital in Digital worlds.

Also, electrical noises, power supply ripples, temperature coefficients that related to a component precision on the scale of it operated temperatures can induce errors as an inflated values (read values + precision tolerances) then rounded down or up. This is why many DAC and AMP sound better warmed up than cold.

yes, sound is subjective. However, the differences is real, and that is why there are precision tolerances, phase noises, temperature co efficient.....those electrical engineerings parameters. All of those would contribute into processing of Binary decoding.

The reason why an SACD is better than PCM CD, was from all of this point. An already determined and no longer able to be further tempered with, directly from a studio to your room. This is a superior Digital medium....ai would take an extreme example that with PCM, applying digital EQ is possible, while DSD, you can not use digital EQ. Digital, DSD, is as close as Analog as possible. (This is exactly how the Studio want you to hear)...Then again DSD64 is the least of 2X oversampling. Which would repeat again that Non Over Sampling is less accurate or resistors ladder is less accurate. However, there are techniques to overcome that, which would also require different DSP and Errors correction techniques, which also alternate the sound performances


My points again, nothing is better, it is just how you prefer to enjoy it, the moment Human record and reproduce artificially, nothing is original. It is just that the real world is Analog, and Recording microphones is a Prime examples of that.

Many people strong point is that Analog throughoutly such as AAA will have a more natural presentation, which is true, because....the real world is analog. Why I stated Analog is the best here is because there are no algorithms errors that can take places here. Unlike the digital. Again, analogs have it own down fall, but all of that is still Analog itself.

the moment we assign quantitative values for something that is Infinite (sin waves) we already taken a further steps from being Artificially recording-reproducing to an additional Artificially construct and reconstruction of the original values

 

[SoundAndMotion]

@Whitigir
Understanding the technical details of the various recording and playback methods* is not required for listening to and enjoying music using any of the methods you own.
Happy listening!


* but it sure helps when trying to make technical arguments. Your lack of understanding is hampering you. The Xiph (aka Monty) video @pinnahertz linked above (as have others) is excellent! I echo the others in recommending it

 

[bigshot]

Yes I know about and still studying digital and it processing, algorithms etc....but allow me to not open a can of worms on how it Effect sounds. You can find Jussi LashKo quote, the creator of HQplayer, and also Sabre engineers, AKM engineers about how digital filters alternate the sound performances I don’t make up stuff, FYI

I know you aren't making this up. The problem is that you are getting your information from people who are trying to sell something to you. They are going to skew and cherry pick to make it look like the other guy's product sucks. It's a lie with the face of a smile.

The reason I told you to google Nyquist is because you have fundamental misunderstandings about the way digital audio works. If you had googled, the wikipedia page would have come up and you would have gotten this sentence right at the top of the article...

In the field of digital signal processing, the sampling theorem is a fundamental bridge between continuous-time signals and discrete-time signals. It establishes a sufficient condition for a sample rate that permits a discrete sequence of samples to capture all the information from a continuous-time signal of finite bandwidth.

SUFFICIENT SAMPLE RATE PERMITS THE SEQUENCE OF SAMPLES TO CAPTURE ALL THE INFORMATION FROM CONTINUOUS TIME SIGNAL.

Analogue is continuous, and within the bandwidth of human hearing 16/44.1 is capable of capturing ALL of the information. It can PERFECTLY reproduce every sound you can hear... no noise, no distortion... perfect reproduction. Better than the best analogue format that exists, because it is perfect within the range that human ears can hear.

This is just the first lesson. There is a lot more to learn that will help you build a kick ass sound system. You just need to get your information from engineers, not stereo salesmen and marketing departments. The problem is that engineers sometimes like to speak more about the molehills than the mountains. You'll have to weed out the minutia and focus on the simplest stuff to learn first.

I'd suggest keeping a pad of paper next to you and write down technical terms you don't fully understand. Then when you have time, google and try to figure out what it means, how it works, and how much of it human ears can hear. If you notice, the wikipedia quote has already done that with clickable links to terms. Check out the links there and the videos in my sig file. They are a great place to start.

It would also help if you didn't automatically argue with people who might know more about this than you do. Take your visit to Sound Science as an opportunity to learn, not a reason to argue. Also realize that a lot of people here have been conditioned to attack by idiots who come in here to argue. I am giving you the benefit of the doubt and I am trying to help you learn.

That's my advice.

By the way, I remember you popping in here a couple of times in the past to troll us. I'm giving you one chance here to prove you are here for the right reason. Answer carefully, because you are being judged.

 

[gregorio]

[1] That isn’t analog, that is Morse Code, and with a messed up timing, this info is meaningless, [1a] these timing have to be so precise as much as femto seconds.

1. Why do you think morse code existed? If the result were meaningless it would never have been used, let alone widely used! Morse code was used because it resulted in pretty much perfect "meaning", while analog over the same distances really was meaningless!
1a. This is the sound SCIENCE forum, so please present some reliable evidence to support your assertion. And, analog timing errors are more that a trillion times greater than "femto seconds", how does that make analog better?

[1] Errors within analog systems consisting physical errors and mechanicals errors.
[2] The errors that happening in digital is digital errors, electrical signals pulsation errors, and mechanical errors.

1. And magnetic errors, or are you talking about very early analogue, before tape was invented?
2. True but the whole point is that those digital errors are orders of magnitude less than the additional mechanical, electrical and magnetic errors required with analogue!

[1] Keeping this signals and replicating it = the best representation of the original.
[2] Yes noises maybe added, but the world is analog, and there is noises everywhere even in vacuum spaces ...
[2a] Many people strong point is that Analog throughoutly such as AAA will have a more natural presentation, which is true, because....the real world is analog.

1. No it's not, because you can't "keep the signals" accurately with analog recording.

2. Huh, the real world is analog? I understand that you're still studying digital but you seem to have skipped over lesson number 1, the real world is NOT analog, it's acoustic! That's why it's called analog, because it is NOT the real world, it's "analogous" to the real world! And, good luck trying to get sound to travel through a vacuum!
2a. "Many people strong point" is therefore nonsense, because they obviously don't even know what the real world is.

The reason why an SACD is better than PCM CD, was from all of this point. An already determined and no longer able to be further tempered with, directly from a studio to your room. This is a superior Digital medium....ai would take an extreme example that with PCM, applying digital EQ is possible, while DSD, you can not use digital EQ. Digital, DSD, is as close as Analog as possible.

This makes no sense at all. Pretty much all commercial music recordings from the 1950's onwards have EQ and/or other processing and of course, they were all analog recordings. So how is DSD as close to analog as possible, when all analogue recordings have something which DSD does not? It's also nonsense because the vast majority of SACDs have been converted to and from PCM in order to apply that processing and if that's not bad enough, you're contradicting yourself, because SACD is absolutely reliant on noise-shaping (dither).

[1] Why I stated Analog is the best here is because there are no algorithms errors that can take places here. Unlike the digital. Again, analogs have it own down fall, but all of that is still Analog itself.
[2] the moment we assign quantitative values for something that is Infinite (sin waves) we already taken a further steps from being Artificially recording-reproducing to an additional Artificially construct and reconstruction of the original values
[3] I don’t make up stuff, FYI

1. And that's what you're mistake is. Analog has errors, digital has errors but they're NOT equal, errors due to digital are less (typically at least an order of magnitude less!) than analog and virtually always below or way below audibility, unlike analog! You're completely ignoring the relative scale/magnitude of the errors and then stating that analog must therefore be better because it's the real world, which is utterly false!

2. You say you've read Nyquist but unfortunately you've completely failed to understand it!

3. What practical difference is there between you making-up nonsense and you repeating nonsense that someone else has made-up?

PLEASE read bigshot's post and take his advice.

G

 

[pinnahertz]

Wait, Noise shaping and Noise filtering, Anti Aliasing are not considered DSP ? Those are the majority of Digital Filterings

Sure, but all of that are tweaks to the basic quantization process, which works without them.

Yes, there are pulsation errors as these codes are read and calculated with the timing by the algorithms and the Digital signal processing techniques. A values given would have to be rounded up or rounded down, if it exceed the values of the resistors to determine the Pulses. That is why R2R and NOS are the least accurate ways to process digital in Digital worlds.

Forgive my ignorance, but I'm unable to find a reference to the term "pulsation error" as it applies to quantization of digital audio. I suspect it doesn't exist outside of marketing, but please link up a reference if you have it.

Samples don't need to be rounded up or down, resistor values in a DAC have nothing to do with quantization. DSP algorithms are not required for quantization to take place, they are all add-ons to the process, and are done at higher bit depth so there is no rounding required. Words are simply truncated.

Also, electrical noises, power supply ripples, temperature coefficients that related to a component precision on the scale of it operated temperatures can induce errors as an inflated values (read values + precision tolerances) then rounded down or up. This is why many DAC and AMP sound better warmed up than cold.

Poor designs exist in all technology, but they aren't limits to the technology itself. There should be no electrical noises in an ADC or DAC system that affect its performance. Power supply ripple is something that has been taken care of for over 75 years, it simply should not, and does not exist in any properly designed and operating equipment, analog or digital. Here's a reality check: a garden-variety, nothing special precision R2R ladder has a tempco of 100ppm/degreeC. That means changing it's temperature by 25 degrees C (that would be a LOT), moves R valules by 0.25%, which is less than the total spec of the unit. If they all drift the same direction, which is highly likely, no additional errors are produced. None.

Once again, a little knowlege filtered through marketing creates disinformation, which then influences belief systems skewed to favor the desired technology.

yes, sound is subjective. However, the differences is real, and that is why there are precision tolerances, phase noises, temperature coefficient.....those electrical engineerings parameters. All of those would contribute into processing of Binary decoding.

But all of those same principles influence analog circuits to a far greater degree! Just because it's analog doesn't get you around the effects of tolerances or temperature induced drift at all, in fact, many more parameters change because in the digital world, all of this has been stabilized. Take your favorite class A amplifier of any time, turn it on and measure distortion. Then wait and hour and measure again. It's not stable at all! Noise will also change.

I won't dispute the different technologies sound different, I will argue in favor of a digital system's ability to replicate the input signal over any analog system on earth.

The reason why an SACD is better than PCM CD, was from all of this point. An already determined and no longer able to be further tempered with, directly from a studio to your room. This is a superior Digital medium....ai would take an extreme example that with PCM, applying digital EQ is possible, while DSD, you can not use digital EQ. Digital, DSD, is as close as Analog as possible. (This is exactly how the Studio want you to hear)...Then again DSD64 is the least of 2X oversampling. Which would repeat again that Non Over Sampling is less accurate or resistors ladder is less accurate. However, there are techniques to overcome that, which would also require different DSP and Errors correction techniques, which also alternate the sound performances

I have no time to debate this other than to say there exists confirmed evidence that this is completely wrong. I'll post a link later, working now.

My points again, nothing is better, it is just how you prefer to enjoy it, the moment Human record and reproduce artificially, nothing is original. It is just that the real world is Analog, and Recording microphones is a Prime examples of that.

Many people strong point is that Analog throughoutly such as AAA will have a more natural presentation, which is true, because....the real world is analog. Why I stated Analog is the best here is because there are no algorithms errors that can take places here. Unlike the digital. Again, analogs have it own down fall, but all of that is still Analog itself.

the moment we assign quantitative values for something that is Infinite (sin waves) we already taken a further steps from being Artificially recording-reproducing to an additional Artificially construct and reconstruction of the original values

Again, sorry, I just don't have the time to get into this other than to say I disagree. The concept that the world is analog therefore analog recording is better is, in my opinion, nonsense. I dont' come by my opinion casually, it's the result of 5 decades of work in the field, attempting to extract the last fraction of a dB of quality out of analog recording systems. I know the limitations and the flaws extremely well. And they can sound very good, but they do not replicate the input signal as well as even rudimentary digital systems. It has nothing to do with the world being analog.

 

[nanaholic]

Is there any point in buying modern day, 21st century pressings of any modern records?

Taking this back to the original topic, really the tl;dr is even if a vinyl is "DDD", there's still a chance that it will sound different to the CD due to the mastering technique difference between the medium, and thus there's still a point to buying vinyl pressings.

As an engineer whom had studied the basics of digital signal processing I'm under no illusion of the BS that vinyl is technically superior because it is "all analogue" or whatever, but rather the fact that there's a not-insignificant chance the mastering is different to the CD which then leads to an objectively better sounding final product, and that may or may not be intentional on the part of the production. Just the other day I bought a new release vinyl of some modern jpop song, I thought it sounded wonderful compared to the original CD release. Having a pretty good idea of why this is so but wanted to confirm my guess I ripped the vinyl, did some gain adjustments and compared the signal to the original CD which was released in 2007 - and sure enough the CD release was mixed way too hot to the point of having clipping in almost all the chorus portions of the song so it became really shout-y, and has almost no dynamic range to speak of. It wasn't because "vinyl is a better sounding medium than CD", but "someone messed up the mastering for the CD pretty badly in the first place", and we know who to put some of the blame of this to (the loudness war leading to bad mastering practices that became prolific in the industry), which means the issue is the thing which sits between the chair and the mastering control panel ie the human making all the adjustments, which is the part which us enthusiasts have nearly absolutely no control or window to look into when we buy these products.

The irony is that the merits, robustness and convenience of the digital production method and the digital medium allows for these types of bad practices to sneak into the final product without much consequences (eg you can clip the signal to hell and the worse it can do is make our ears bleed when it comes out of the speakers/earphones, but it won't really damage the equipment), while the limitation of vinyl actually forms a better framework to keep good mastering practices in place.

 

[old tech]

The irony is that the merits, robustness and convenience of the digital production method and the digital medium allows for these types of bad practices to sneak into the final product without much consequences (eg you can clip the signal to hell and the worse it can do is make our ears bleed when it comes out of the speakers/earphones, but it won't really damage the equipment), while the limitation of vinyl actually forms a better framework to keep good mastering practices in place.

That is the point The Sonic Truth and I were discussing a while back. Digital recordings are superior to analog recordings (it would be a rare classical music audiophile that would disagree with that statement) but the superiority can be abused in production.

I agree that a record can sound different to a CD or any other digital format even when the same digital master is used (or for that matter, if the same analog master is used for both). The example you gave is a more obvious example when the production master or lacquer for the LP has been tweaked. However, most modern LPs use the same final master as the CD (or share the same 24/96 master), perhaps with lower volume levels on the LP to deal with any hypercompression.

They can sound remarkably similar. The 1978 Japan Pro Use LP and the 1983 Japan Black Triangle Dark Side of the Moon CD is an example where they both used the same analog 15 ips master, with the CD being a flat transfer. I have both and they sound almost identical - the LP being slightly different due to the sound character of my TT/cart set up and, if listening critically, the IGD and slightly lower dynamics of the LP, particularly towards the end of the sides (though back in the day when I had a laser turntable even that IGD difference disappeared). That LP btw, is one of the quietest, and perfectly flat and centred LPs I've come across.

I can't quite get my head around a DDD record. Surely the most if can be is DDA, as the record is always analog.

Lastly, that argument that analog is better because the world is analog is often thrown about (without any supporting evidence) even by some people who should know better. If that was the case then two cans held together by a tight string should sound better than a reel to reel or LP as it is a more perfect analog (less processing and less change of energy states). The other variation to this is that "our ears are analog" - well yes but our brains are not. Our neurons are more like digital, ie either fire or they don't, similar to 1s and 0s. However, some on the digital side take this argument too far as well.

 

[old tech]

Wait, Noise shaping and Noise filtering, Anti Aliasing are not considered DSP ? Those are the majority of Digital Filterings

Assuming you may be interested in real world objective measurements, this article by Archimago may be relevant.
http://archimago.blogspot.com/2017/06/musings-measurement-thoughts-on-vinyl.html

 

[nanaholic]

I agree that a record can sound different to a CD or any other digital format even when the same digital master is used (or for that matter, if the same analog master is used for both). The example you gave is a more obvious example when the production master or lacquer for the LP has been tweaked. However, most modern LPs use the same final master as the CD (or share the same 24/96 master), perhaps with lower volume levels on the LP to deal with any hypercompression.

In a perfect world you would be right, but real world experience says pretty much otherwise. And I discovered this from no other place but the world of Hi-res downloads vs CDs.

Going from pure production and engineering logic, you would think when a studio master is made (using studio quality bit depth and sampling rate ie 24/48 or higher, which automatically qualifies it as hi-res), the hi-res is just that final master as is and the CD would just be a downsample of the same studio master, but otherwise identical in the all important frequency range of 0-22khz, no? But absolutely not a single one of the hi-res release of the exact same CD album I have bought in the past 4 years are anything alike in their gain and dynamic range.

It seems there is some agreed upon practice where even when using the exact same master, during final mastering the engineer MUST tweak *something*, and usually for the humble CD that means throwing the prestine studio master into a compressor and gain booster to make it super squashed and super loud (just a hair below clipping), and that simple act is what ruins the majority of the CD releases today, even if the same studio master is used. And I sincerely believe this is the exact same case for CD and vinyl releases - the studio master is still (required?) tweaked in some ways and that tweaking makes or breaks the release, no matter which medium you are talking about, and I don't believe any label releases the very original complete master for the consumer to buy, ever.

So even when they claim there's three stages to all of this, I think in reality there's actually a fourth step, and it's more like

1) Recording
2) Mixing
3) Mastering -> shelf and store this version (nobody gets to buy this - this is purely a production archive version)
4) Tweaked the archive version further for which ever target release medium (CD, vinyl, hi-res/normal digital downloads, streaming, radio etc)

It's that 4th step which I care about the most these days, hence why I end up buying multiple version of the same album on different mediums.

I can't quite get my head around a DDD record. Surely the most if can be is DDA, as the record is always analog.

I'm assuming when the initial recording is instantly digitized as it is being recorded from the microphone input and never goes on any analogue magnetic tape, then technically it would be a full DDD production path.

You can find stories on the internet where recording studios is using magnetic tapes for the initial recording stage and brags about it being "better".

 

[bigshot]

Hot mastering is more of a problem with pop music aimed at people who listen to music on their phone. It's intended for people who put their music on random shuffle. Every song in the playlist needs to come up at the same volume. The levels can't change throughout the album because you can't predict what order the songs will come up in. Hot mastering is almost unheard of in classical and jazz. There's quite a bit of pop music that isn't hot mastered too- you just have to look for formats marketed to audiophiles.

 

[nanaholic]

There's quite a bit of pop music that isn't hot mastered too- you just have to look for formats marketed to audiophiles.

And that answers the original question - because vinyl is a audiophile format so they are usually not mixed as hot as the CD and thus it is still worth buying.

Also surprisingly download and streaming services like Apple and Youtube are providing much stricter mixing guidelines so they don't get too hot, it's not rare that you can find iTunes Store downloads which sounds better than the CD version as well, even though they are not "audiophile" targeting being lossy and all that.

 

[bigshot]

Blu-ray audio, SACD and well remastered CDs can be audiophile formats too, but without the limitations of LPs.

 

[old tech]

And that answers the original question - because vinyl is a audiophile format so they are usually not mixed as hot as the CD and thus it is still worth buying.

Also surprisingly download and streaming services like Apple and Youtube are providing much stricter mixing guidelines so they don't get too hot, it's not rare that you can find iTunes Store downloads which sounds better than the CD version as well, even though they are not "audiophile" targeting being lossy and all that.

Well that had me choking on my breakfast... what makes you think vinyl is an audiophile format or that it is mastered that way because it is considered an audiophile format? They are usually mixed not as hot as CD because the vinyl format is an inferior format to the CD, ie it simply is incapable of being mastered that way. CDs, or 16/44 have higher resolution and dynamic range than vinyl and a much higher fidelity. There are (or were) audiophile CD masterings (as there are with vinyl, and cassette tape in the past), eg MFSL, DCC, Decca along with many regular releases by bands that took a strong interest in high fidelity that often sounded better than the audiophile labels (like Dire Straits as one of thousands). Sound is subjective but to most ears these CDs easily sound better than their 'audiophile' vinyl equivalents, and certainly have higher fidelity in any way it can be measured.

 

[old tech]

Blu-ray audio, SACD and well remastered CDs can be audiophile formats too, but without the limitations of LPs.

And original (ie not-remastered) CDs such as those issued by Decca, Telarc and others. They have higher fidelity than audiophile LPs and cassette tapes of the same albums released by these labels. Take Andre Plevin's Holts the Planets 1987 Telarc audiophile CD and compare it with the Telarc audiophile LP. Not only does the CD have better clarity, separation, instrument timbre it also has the dynamic range to do justice to the orchestral sound.

 

[gregorio]

Just the other day I bought a new release vinyl of some modern jpop song, I thought it sounded wonderful compared to the original CD release. Having a pretty good idea of why this is so but wanted to confirm my guess I ripped the vinyl, did some gain adjustments and compared the signal to the original CD which was released in 2007 - and sure enough the CD release was mixed way too hot to the point of having clipping in almost all the chorus portions of the song so it became really shout-y, and has almost no dynamic range to speak of. It wasn't because "vinyl is a better sounding medium than CD", but "someone messed up the mastering for the CD pretty badly in the first place", and we know who to put some of the blame of this to (the loudness war leading to bad mastering practices that became prolific in the industry)

We have to be careful about jumping to conclusions, how do you know that the chorus portions of the song are not supposed/intended to sound "shout-y"? Maybe that "shout-y" chorus actually sounds better in some listening conditions, listening conditions which can never exist with vinyl but are common with CD or maybe, the CD has indeed been "messed-up", an unavoidable consequence of achieving a loudness level demanded by the label or artists.

[1] The example you gave is a more obvious example when the production master or lacquer for the LP has been tweaked. However, most modern LPs use the same final master as the CD (or share the same 24/96 master), perhaps with lower volume levels on the LP to deal with any hypercompression.
[2] I can't quite get my head around a DDD record. Surely the most if can be is DDA, as the record is always analog.

1. Generally I would agree, the master (or technically "pre-master") is the same, *sometimes* tweaks are required for vinyl production, for example more mono and more compressed bass. These tweaks are not performed in the mastering studio by the mastering engineer (again, technically pre-mastering studio/engineer) but at the pressing plant's lab, by a cutting engineer. The creation of the glass master for CD production/duplication doesn't require any tweaks or a cutting engineer. However, when I say "generally", I'm not referring to the fact there are some exceptions to this rule but to the fact that the Japanese market is often a law unto itself and so your statement maybe significantly less true in this particular case.

2. Really, even after I explained it to you just a few days earlier? To recap post #5507 in the 24 vs 16bit thread: The first "A" or "D" represents the original tracking/recording format (Analogue or Digital) of the musicians' performance, the second "A" or "D" represents the recording format of the mix and the third represents the recording format of the (pre) master. So it's entirely possible that an LP can be DDD, and the vast majority of LPs of the last couple of decades or more probably are.

[1] Going from pure production and engineering logic, you would think when a studio master is made (using studio quality bit depth and sampling rate ie 24/48 or higher, which automatically qualifies it as hi-res), the hi-res is just that final master as is and the CD would just be a downsample of the same studio master, but otherwise identical in the all important frequency range of 0-22khz, no? But absolutely not a single one of the hi-res release of the exact same CD album I have bought in the past 4 years are anything alike in their gain and dynamic range.
It seems there is some agreed upon practice where even when using the exact same master, during final mastering the engineer MUST tweak *something*, and usually for the humble CD that means throwing the prestine studio master into a compressor and gain booster to make it super squashed and super loud (just a hair below clipping), and that simple act is what ruins the majority of the CD releases today, even if the same studio master is used.
[2] Also surprisingly download and streaming services like Apple and Youtube are providing much stricter mixing guidelines so they don't get too hot ...

1. That's not entirely how it works. The studio master would be at 64bit (or a decade+ ago, 32bit) and whatever sample rate (44, 96 or 192), from this virtual master the release versions are recorded. IE. Typically truncated to 24bit or down-sampled and noise-shaped dithered down to CD. This is not always the case though, sometimes an additional virtual master is made specifically for CD, with additional compression. In fact, somewhere in the middle of the "24bit vs 16bit myth exploded" thread a representative of Linn Records or HDTracks (I don't remember which) stated that they specifically require their engineers to apply additional compression to all their CD masters. The reason given was that the CD version was likely to be used (or ripped and used) under listening conditions where additional compression is better/advantageous. While this is (or was) a valid/true reason, the NOT coincidental consequence was that the CD version would be audibly different from the 24bit version and probably sound worse under audiophile listening conditions, thereby justifying it's significantly higher price. This is also sometimes the case between the CD and DSD layers on SACDs, though not always.

2. Not really. I can't remember off the top of my head if Apple's (previously MFiT) guidelines even mentions compression but if it does, at most it's a recommendation rather than a guideline, same with Youtube. In practice, when using "sound check" or replaygain, there's no advantage (only disadvantages) to applying excessive amounts of compression but this doesn't yet appear to have changed the situation greatly, by those who commission masters (labels/artists). This is quite a complex issue and there's no consensus between the various companies/organisations.

G