24bit vs 16bit, the myth exploded!
Jan 7, 2020 at 1:51 PM Post #5,536 of 7,175
Well, my 'argument' for higher specs is from a production(including recording sessions) context, not from one of the end-listener formats. I apologize if my argument came off otherwise - a la audiophile, etc.
It doesn't matter. The recording process was 16/44.1/48/50 for the first two decades of digital audio, and it wasn't a limiting factor. The practical limiting factors in "resolution" are all outside of quantization, and way outside quantization at 24 bits. Until digital desks with adequate DSP arrived, most post was done in the analog world, then mixed back to 16/44.1.

The one quality limiter of all PCM in the early days (technically still outside of quantization) was analog anti-aliasing and reconstruction filters, the earliest of which suffered from nonlinearities in the cutoff zone which resulted in intermodulation products being folded down into the mid-audible band. Once oversampling filters arrived, the problem was mostly mitigated. Increasing sampling frequency was initially viewed as the solution, but all that could do was relocate the problem. The real work was on the filters, and there was a cafe of exotic retrofits for every major recording device. All were compromises, trading one filter quality for another, nobody every made the perfect analog filter. You could reduce intermod, but then raise aliasing, for example. Until we got to oversampling digital filters, that was a rough spot.

24 bits couldn't help the intermod issue then, and still wouldn't.

Again, 24bits does not increase resolution or reduce distortion, it lowers quantization noise only. But system noise is already higher than LSB jitter in 24 bits by many, many dB, so it's a wash.

Oh yeah, 24 bits does provide one very significant advantage that all of us in the industry recognize: higher numbers are better, and more bits makes everyone feel warmer and fuzzier.
 
Jan 7, 2020 at 3:58 PM Post #5,538 of 7,175
Thanks for the detailed reply! I guess I'll just stick with 48khz, though I don't really hear a difference between 48 and 44.1, but just to be safe) Also most films and many games use 48, and it's easier on the system (if it's old, at least) to keep playing the same sample rate.
Here's some interesting and detailed material on the matter. https://forums.stevehoffman.tv/threads/longish-why-i-prefer-48-khz-over-44-1-khz.347515/

48 kHz is totally fine as a "fixed" samplerate for all material. The "optimal" samplerate would be something like 56 kHz (because it allows somewhat relaxed anti-aliasing and reconstruction filters). If we can live with steep filters 44.1 kHz is just enough.
 
Jan 7, 2020 at 4:27 PM Post #5,539 of 7,175
They'd sound exactly the same as what you have now. None of that is demanding material at all, and none comes even close to needing 16 bits.

Nice to see things don't change here. All the time I've been away, and it's the same arguments by the same people ignoring the same experts.

Turns out the 16 bit vs 24 bit myth can't be exploded… …it is too persistant! :dizzy_face:

Surprised to see you back here. :relieved:
 
Jan 7, 2020 at 4:32 PM Post #5,540 of 7,175
48 kHz is totally fine as a "fixed" samplerate for all material. The "optimal" samplerate would be something like 56 kHz (because it allows somewhat relaxed anti-aliasing and reconstruction filters). If we can live with steep filters 44.1 kHz is just enough.
I wouldn't classify filter quality based on cutoff slope alone. Much more to it than just that. But yeah, higher sampling mitigates the filter. I just don't think it makes any audible difference if the filters are clean to begin with.
 
Jan 7, 2020 at 5:04 PM Post #5,542 of 7,175
1 --- I wouldn't classify filter quality based on cutoff slope alone.

2 --- I just don't think it makes any audible difference if the filters are clean to begin with.

1 --- Of course not. What I mean here is higher sample rate gives more freedom for the shape of the cutoff slope and at around 56 kHz sample rate the benefits of this freedom are pretty much exhausted.

2 --- Well, I can hear very minor changes in sound (spatial width) when changing the reconstruction filter of my NAD C565BEE CD-player, It's kind of like moving the speakers a few inches closer or away from each other. The standard filter sounds most narrow while the other options gives a little bit wider soundstage of various degree. I also think I can just hear a tiny change in the amount of the highest frequencies, but this might be placebo or just me misinterpreting the change in spatial width as spectral change. All filters give equally high fidelity in my opinion, just a little bit different ways (moving the speakers an inch or two makes similar changes). So, in my opinion at 44.1 kHz the type of reconstruction filter does make an audible difference, but does it matter? Not really if you ask me. If your gear has only one fixed filter your ears will get used to it.
 
Jan 7, 2020 at 5:11 PM Post #5,543 of 7,175
Ha ha! It's only the people that are persistent.
Thanks.

And it's not only this issue. The internet is about insisting one's beliefs ad nauseum and I am one to blame. :rolling_eyes:
I'm trying to learn to become a happier and better person, but it's a long process...
 
Jan 7, 2020 at 6:19 PM Post #5,544 of 7,175
1. As I've mentioned, "those later remastered CDs" were not the product of 24bit processing, as there's never been 24bit processing used in commercial music production to my knowledge. It was the product of 32bit float or 48bit fixed processing.

The point is the same though, 32bit or 48 bit float (or even 24 bit) would enable far more processing of the recording, pushing the sound wars into the stratosphere. What I was trying to say is that there was a limit to how much compression etc when processing power was limited to analog equipment or 16 bit digital. In other words, the greater processing power could and is (in many cases) abused. CDs with DR2 for anyone?

2. To clarify, with the exception of some/a few classical recordings, the process was: Recording on multi-track tape, constantly replaying that tape out to an analogue desk (with "outboard" gear) where it was mixed. When all the desk's (and outboard gear's) parameters we adjusted to produce the desired mix, the result was "bounced" (recorded) back down to tape. There was no such thing as an analog workstation. Digital recording did not change this workflow at all, it was still multi-track tape, constantly replayed out to an analogue desk, etc. The only difference was that the multi-track tape recorded digital audio (and then replayed it through the recorder's DACs) rather than analogue audio. The only workflow change was in editing, as digital audio tape couldn't be spliced and was a rather more involved and time consuming process. Digital workstations started to be used in the mid 1990s but not as workstations, they we're used for editing because they massively reduced the editing time, as well at it being more accurate and non-destructive (and bit depth is irrelevant to editing as there's no processing involved). It wasn't until the very end of the 1990's that they started being used as workstations (IE. For recording, editing and mixing), the first No.1 done this way was Ricky Martin's "Living La Vida Loca" in 1999, although outboard analogue gear was still employed and the mastering was still analogue. Fully ITB (In The Box, no analogue outboard gear) didn't start really taking over in the commercial music world until the mid 2000's, quite a few years after 32bit or 48bit fixed mix environments/processing was standard, with mastering being the last bastion to hold out for a few years more.

Baring all the above in mind, I don't really understand what you mean by "flat transfers" or what TheSonicTruth is trying to say with his response?

G

The recording process you describe still provides limitations on the degree of processing, ala the soundness wars. However, I think we were mainly alluding to the reissue of back catalogs when CD sales started to take off in the early to mid 80s. Of course I am no expert but back in the day a close friend of mine worked at Alberts and Disctronics and many of these releases were a flat transfer of whatever analog production tape was available. Depending on the quality of the tape (some were great and others quite poor (condition and/or generation wise) which I think accounts for the large variability in sound quality of many 80s back catalogue CDs. What I (and Sonic Truth, I think) were getting at is that many or most of these flat transfer CDs subjectively sound better than their later remasters simply because they were a flat transfer without any futzing or attempt to make them louder and more compressed/limited.[/QUOTE]
 
Jan 7, 2020 at 6:40 PM Post #5,545 of 7,175
Whether or not a remastered album sounds better or not depends more on how the processing is applied, than it does whether processing has been applied at all. Bad remastering will result in a worse sounding album. Good processing would result in a better sounding album. Compression and noise reduction have existed since the late 1930s, and I have 78s that have been over compressed and over scrubbed. It's nothing new. However, the tools used to compress and reduce noise have gotten much better since the 1980s. Sure, they can still be used to blunt and mangle music. But used correctly and with good taste, they can vastly improve old analogue masters. Mixing and mastering is a subjective process, and it serves different purposes with different formats and audiences. I've said this a million times to Sonic Truth (and he just doesn't hear what I am saying)... There is good remastering and bad remastering. You can't generalize that most remastering is bad and most 80s CDs are better. There are too many examples of just the opposite being true. You have to judge albums on a case by case basis. The words "24 Bit Remastering" are nothing more than sales pitch. The phrase doesn't indicate whether the album sounds better, worse or exactly the same. You have to listen to it to find that out.
 
Jan 7, 2020 at 6:46 PM Post #5,546 of 7,175
What I (and Sonic Truth, I think) were getting at is that many or most of these
flat transfer CDs subjectively sound better than their later remasters simply
because they were a flat transfer without any futzing or attempt to make them
louder and more compressed/limited.


THANK YOU, ot! Glad I am not alone in this regard, nor completely nuts, lol!

Not always "better", in the case of those first-issue CDs, but: generations closer to the sound of the original musical intent of the given album and its artists. As I keep repeating on here, alas to deaf ears, but "better" is purely subjective, from a audio perspective. Don't get me wrong: take two identical make/model# stereo speakers, and the one with intact driver surrounds will definitely sound better than the one with torn or rotted away surrounds. Common sense! But that's not what we're on about, here.

Back to point: Sadly, many of those first-issue era CDs did suffer from some of the distortions Gregorio mentioned a page back, post #5531 in this conversation, something all the recording bits and stratospheric sampling rates on Earth would have hardly put a dent in, noise-wise, but still, I prefer them to those reissues from during the 'Remaster-Mania' Era(Late 1990s-2000s).
 
Last edited:
Jan 7, 2020 at 7:08 PM Post #5,547 of 7,175
I've got lots of CDs from the 1990s and 2000s that sound better than they ever sounded before. I don't know how you can generalize unless you're specifically speaking about albums that were poorly remastered. There were plenty of those too. They were pumping out lots of titles back then... some good, some bad.
 
Last edited:
Jan 7, 2020 at 7:23 PM Post #5,548 of 7,175
Whether or not a remastered album sounds better or not depends more on how the processing is applied, than it does whether processing has been applied at all. Bad remastering will result in a worse sounding album. Good processing would result in a better sounding album. Compression and noise reduction have existed since the late 1930s, and I have 78s that have been over compressed and over scrubbed. It's nothing new. However, the tools used to compress and reduce noise have gotten much better since the 1980s. Sure, they can still be used to blunt and mangle music. But used correctly and with good taste, they can vastly improve old analogue masters. Mixing and mastering is a subjective process, and it serves different purposes with different formats and audiences. I've said this a million times to Sonic Truth (and he just doesn't hear what I am saying)... There is good remastering and bad remastering. You can't generalize that most remastering is bad and most 80s CDs are better. There are too many examples of just the opposite being true. You have to judge albums on a case by case basis. The words "24 Bit Remastering" are nothing more than sales pitch. The phrase doesn't indicate whether the album sounds better, worse or exactly the same. You have to listen to it to find that out.
I don't think we are on different wavelengths... I agree with what you say. I'm not doubting that compression, NR, etc have always existed (and for good reasons), the point is that the original production masters would already have been mastered with it (ie compression, EQ and other 'tricks') that the original producer/artist wanted. Sure, there are many remasters that do (subjectively) sound better and often they are the albums that were remastered with the aim of improving the original, mitigating their flaws etc rather than just making it loud and brickwalled because that is a modern trend. Of course it is case by case but generally, mastering has got hotter and more limited/brickwalled since the early 1990s. Hi res processing is neither good or bad, it depends on how it is used. The part we were debating was whether (on balance) the introduction of "hi res" processing has mainly resulted in improvements or a decline in sound quality for the 5% of the population that value sound quality.
 
Last edited:
Jan 7, 2020 at 7:31 PM Post #5,549 of 7,175
The point is the same though, 32bit or 48 bit float (or even 24 bit) would enable far more processing of the recording, pushing the sound wars into the stratosphere. What I was trying to say is that there was a limit to how much compression etc when processing power was limited to analog equipment or 16 bit digital. In other words, the greater processing power could and is (in many cases) abused. CDs with DR2 for anyone?
Hmm.. Well let me relate something about processing that may shed some light. A good part of my career has been spend in broadcasting where there is a very aggressive version of remastering processing going on in the final processor before transmission. The idea is quite old, but the tools got pretty intense in the late 1970s, early 1980s, all analog of course. The goal was to be loud, subjectively clean (the two are opposite vectors), and not go past legal modulation limits. The degree varied by music type and format, but the general goal is always true. The loudness war on-air is older than it is in recorded music. So the analog tools got pretty fancy. Then suddenly, we had digital processing, with tons more flexibility, stability, and the ability to save and recall massive settings. You know what the goal was then? To make that DSP based megaprocessor sound as good as it's analog predecessor. And it took almost a decade to come even close to that goal. Now we have the latest units with massive DSP engines, we can build extremely complex processors, and we have the ability to model nonlinear systems. Yeah, some of the new digital processors are amazing, but I'll tell you, first hand, it's not easy to reach the goal with those tools. The flexibility works against you. In fact, there are some rather prominent products that are actually pretty far behind what we did analog 30 years ago, and not because we can't, it's because everybody want's to tweak the latest tools to push for louder and cleaner. The two are still opposite vectors. Nothing has changed but the number of ways and degree to which you can mangle the signal.

So, 48 bit float would enable more processing? I don't disagree, but the result would be no better than what you could do with fewer (or zero) bits because it's about the algorithm first and foremost, not about bit depth at all. The most you can hope to gain in the digital realm is for tight and accurate peak control, and even then only if you fully include the entire system in the algorithm, even if it is predictive. All of this is because 90% of all dynamics processing happens above -15dBFS where we have all bits active already. Remember, higher bit depth only lowers quantization noise, it does not increase resolution. Dynamics processing works in the area farthest from the noise floor by nature.

There is no "stratosphere" here. There's no way to magically raise 0dBFS 3dB. In the end, you can only do so much before the damage is just too great to live with. Where that limit is changes with the individual making the judgment, not the DSP or bit depth.

So I'd have to say high bit depth processing doesn't gain you anything when it comes to loudness-war processing. The ability perform effective loudness processing is not contingent on word bit depth, it's entirely dependent on how the specific process works and how it is employed.
 
Jan 7, 2020 at 7:32 PM Post #5,550 of 7,175
Whether or not a remastered album sounds better or not dependsmore on how the processing is
applied, than it does whether processing has been applied at all. Bad remastering will result
in a worse sounding album. Good processing would result in a better sounding album. Compression
and noise reduction have existed since the late1930s, and I have 78s that have been over compressed
and over scrubbed. It's nothing new. However, the tools used to compress and reduce noise have
gotten much better since the1980s. Sure, they can still be used to blunt and mangle music. But
used correctly and with good taste, they can vastly improve old analogue masters. Mixing and
mastering is a subjective process, and it serves different purposes with different formats and
audiences. I've said this a million times to Sonic Truth (and he just doesn't hear what I am saying)...
There is good remastering and bad remastering. You can't generalize that most remastering is bad
and most 80s CDs are better. There are too many examples of just the opposite being true. You
have to judge albums on a case by case basis. The words "24 Bit Remastering" are nothing more
than sales pitch. The phrase doesn't indicate whether the album sounds better, worse or exactly
the same. You have to listen to it to find that out.

But I'm not debating original Vs remaster on a 'which sounds better' basis.

I'm generally opposed to remastered reissues of classic pop genre(rock, rap, country, etc) because the sound of the reissued music has been changed from either what I'm used to it sounding like, or what I remember it sounding like, etc etc. It's not whether the remastered version sounds better or worse, but rather that it sounds different. I'm sorry to anyone whom I might have previously and unintentionally misled, on this point, in the past.
 
Last edited:

Users who are viewing this thread

Back
Top