Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Nice, a personal attack when asked for a technical answer. I bow down to your golden ears. :D

If you'd like to find out how and why this stuff works for real, please come and join us in the Sound Science forum at Head-fi where there are plenty of opportunities to check out your biases with real tests.

not a personal attack, I'm just a too lazy to find all my old notes, then find the relevant ones, but I do remember the gist of it all. And while you cling to your Red Book audio I'll continue yearning for something better, data storage is cheep these days.
 
not a personal attack, I'm just a too lazy to find all my old notes, then find the relevant ones, but I do remember the gist of it all.

What you'd find if you would re-read the notes is that you only get the "perfect" reconstruction of the highs if you low pass filter at half the sample rate AND your low pass filter is impossibly perfect.

The proof is trivial: Assume there is no filter. Now send a 22,050Hz sine wave to the interface at 44.1K samples per second. What do you have? A square wave. What is a square wave? It's the sum of a sine and all of it's odd harmonics

The perfect filter removes all the frequencies above (say) 20KHz which would eliminate all the odd harmonics (as each is well above 20KHz) but any real filter will leave in some of them. In fact a reasonable filter would only cut the third harmonic by about 36dB.
 
What you'd find if you would re-read the notes is that you only get the "perfect" reconstruction of the highs if you low pass filter at half the sample rate AND your low pass filter is impossibly perfect.

don't remember that specifically, but all the rest I'm sure is in there (in my notes). That's actually quite interesting that you'd have to filter right down at half the sample rate, didn't realise red book was THAT far from really doing the job.
 
don't remember that specifically, but all the rest I'm sure is in there (in my notes). That's actually quite interesting that you'd have to filter right down at half the sample rate, didn't realise red book was THAT far from really doing the job.

So did this sound engineering course not cover the Nyquist theorem?
 
So did this sound engineering course not cover the Nyquist theorem?

I remember Nyquist frequency, come to think of it... I must have been 1/2 asleep when I typed that lol, re reading that I feel like an idiot. oh well there goes all my credibility :|
 
I could tell you HOW specifically, but I can't be bothered looking though all my old notes and such from the course I did in sound engineering just to settle a petty argument on the internet, if you really can't hear the difference, then either your ears are failing, or my ears are better than I ever realised

If your notes say this, then your class was like one of those physics 101 classes where the student excesses have notes on them that say things like "ignore the effects of air resistance."

Remember the physics example where they ask how fast a bullet shot straight up will be going after it falls down and impacts the Earth. It is an easy and neat problem if you ignore the real-world effects of air and a very hard problem if you don't.

The case where the signal is perfectly reconstructed is like the bullet problem. It is a simplified example given to students and ignores real problems such as filters not having perfect step functions, harmonic distortion in playback systems and 50 other things.
 
well for the most part it was a practical course, so needing to know the ins and outs of ALL real world effects wasn't a big deal, for example we learnt that temperature and humidity can affect how fast sound travels through air, but not exactly how much or how to calculate them, and to simply assume sound travels at 333m/s
 
I have more than 'decent' equipment. I could buy a nice car with the headphone arrangement I have.

Of course you get fewer samples per wavelength at higher frequencies, but 44.1 still provides enough samples for perfect reproduction to above 20kHz. More samples won't make it more perfect.

The differences you hear between 96k and 44.1k material is because of differences in the mastering.

So, HOW are higher frequencies and transients reproduced better than the already perfect reproduction that 44.1 gives us? Be specific please.

I'm sorry, but that's nonsense. 16/44 sounds nothing like 24/96 or 24/192...unless there's something wrong with your hearing. If 16/44 was good enough, 24/96, 24/192 and DSD wouldn't exist.
 
Here is an article that conflicts with the idea that 24/96 is better then Redbook. Also, the article referrences the BAS (Boston Acoustic Society) blind tests that conclude the same thing.

Personally, I think the mastering that goes into the hirez discs is more important that DSD or 24/96 files. The loudness war, mentioned earlier, has done more damage to the reputation of the CD format than the format itself. I should mention that I own SACDs/DVD-A because it is the only place to get better mastering in some cases.

As for the original question, it is a tough to know because we are talking about mastering. If the original CD is a victim of the loudness war and has limited dynamic range and clipping, I would rather hear a lossy version (assuming it is above 256k) of a better mastered version of the music. I guess my final answer is that it depends on how awful the original master is. It would be nice if we could get the better master in lossless format, but I guess they will hold that out for now and convince us to upgrade later.
 
Why waste time reading a BS article when it's obvious to anyone with a pair of functioning ears that 24/96 is better?
BAS (Boston Acoustic Society) did a thorough study that doesn't support your argument. The difference you are hearing is most likely related to the mastering. Unfortunately, you may only be able to find "audiophile quality" mastering with hirez formats.
 
BAS (Boston Acoustic Society) did a thorough study that doesn't support your argument. The difference you are hearing is most likely related to the mastering. Unfortunately, you may only be able to find "audiophile quality" mastering with hirez formats.

It has absolutely NOTHING to do with mastering. 16/44 is horrible. This is easily heard USING THE SAME MASTER. You should probably get a hearing test.
 
It has absolutely NOTHING to do with mastering. 16/44 is horrible. This is easily heard USING THE SAME MASTER. You should probably get a hearing test.

I have never done a side by side comparison because there has never been a need for it. However, Boston Acoustic Society did a year of blind tests with everyone apparently needing a hearing test:

"[Engineering Report] Claims both published and anecdotal are regularly made for audibly superior sound quality for two-channel audio encoded with longer word lengths and/or at higher sampling rates than the 16-bit/44.1-kHz CD standard. The authors report on a series of double-blind tests comparing the analog output of high-resolution players playing high-resolution recordings with the same signal passed through a 16-bit/44.1-kHz “bottleneck.” The tests were conducted for over a year using different systems and a variety of subjects. The systems included expensive professional monitors and one high-end system with electrostatic loudspeakers and expensive components and cables. The subjects included professional recording engineers, students in a university recording program, and dedicated audiophiles. The test results show that the CD-quality A/D/A loop was undetectable at normal-to-loud listening levels, by any of the subjects, on any of the playback systems. The noise of the CD-quality loop was audible only at very elevated levels."


I own Hi-Rez discs because they sound better, but I am not convinced that the reason is not related to mastering.
 
I have never done a side by side comparison because there has never been a need for it. However, Boston Acoustic Society did a year of blind tests with everyone apparently needing a hearing test:

"[Engineering Report] Claims both published and anecdotal are regularly made for audibly superior sound quality for two-channel audio encoded with longer word lengths and/or at higher sampling rates than the 16-bit/44.1-kHz CD standard. The authors report on a series of double-blind tests comparing the analog output of high-resolution players playing high-resolution recordings with the same signal passed through a 16-bit/44.1-kHz “bottleneck.” The tests were conducted for over a year using different systems and a variety of subjects. The systems included expensive professional monitors and one high-end system with electrostatic loudspeakers and expensive components and cables. The subjects included professional recording engineers, students in a university recording program, and dedicated audiophiles. The test results show that the CD-quality A/D/A loop was undetectable at normal-to-loud listening levels, by any of the subjects, on any of the playback systems. The noise of the CD-quality loop was audible only at very elevated levels."


I own Hi-Rez discs because they sound better, but I am not convinced that the reason is not related to mastering.

read through that again it seems to say that the hi-res audio was passed through a 16bit/ 44.1k bottle neck, therefore proving NOTHING
 
read through that again it seems to say that the hi-res audio was passed through a 16bit/ 44.1k bottle neck, therefore proving NOTHING
They compared true hi-rez to the same hi-rez run through a 16/44.1 bottleneck. It shows that 16/44.1 sounds the same to all of their subjects as the hi-rez using the same source file from the hi-rez master. It is the only way to compare since the hi rez versions have a better master than the CD thanks to the "loudness wars". Read it again.
 
It has absolutely NOTHING to do with mastering. 16/44 is horrible. This is easily heard USING THE SAME MASTER. You should probably get a hearing test.

You serious man? Have you even listened to the difference? If you weren't a member on here since 2008, I'd say you're trolling. I even just tried bouncing my own track as 24/96 and 16/44.1 (recorded in 24/96). I can't hear the difference. I have decent playback equipement, a pretty good ear and know the song pretty well... We record in 24 bit because we don't want to worry about headroom or bother compressing before hitting the DAC.
 
It has absolutely NOTHING to do with mastering. 16/44 is horrible. This is easily heard USING THE SAME MASTER. You should probably get a hearing test.

So I also invite you over to the Head-Fi Sound Science forum where you can test your golden ears for yourself. There is a huge body of evidence that says you are hearing things; almost no evidence whatsoever supporting your position. Come and try some blind tests for yourself.

I suspect you won't though because the difference is SO OBVIOUS. :rolleyes:
 
So I also invite you over to the Head-Fi Sound Science forum where you can test your golden ears for yourself. There is a huge body of evidence that says you are hearing things; almost no evidence whatsoever supporting your position. Come and try some blind tests for yourself.

I suspect you won't though because the difference is SO OBVIOUS. :rolleyes:

In my experience as a life long researcher, incompetence and corruption in the research field is significantly more prevalent than in many other fields. The paired comparison switching identification research methodology being used in these tests is clearly DICTATING the results. That's blatant incompetence on the part of anyone using those methods.

When you develop some results based on a valid monadic non-switching non-identification methodology, let me know. I suspect you won't though, because those results will clearly indicate that 16/44 sucks and that's apparently something you don't want to hear.
 
Last edited:
In my experience as a life long researcher, incompetence and corruption in the research field is significantly more prevalent than in many other fields. The switching research methodology being used in these tests is clearly DICTATING the results. That's blatant incompetence on the part of anyone using those methods.

When you develop some results based on a valid non-switching methodology, let me know. I suspect you won't though, because those results will clearly indicate that 16/44 sucks and that's obviously something you don't want to hear.

You are the one asserting that the methods are flawed; therefore it is up to you to back up your statements.

There is hard science behind my position, with studies to back it up. I have no beef against high-resolution per se, but I do have a problem with an industry that takes an incredible amount of money from people by purposefully distorting data, attempting to sow confusion, and in many cases by using outright lies. At this point everything is leaning towards the fact that 'high rez' is just another way for record companies to sell new versions of the same albums yet again. All of the hard data shows that it makes no audible difference, yet people like you are always willing to chime in with absolute certainty that they can hear something that nobody else can (despite the fact that in over 100 years of audiology research, there has never once been a single discovery of a subject with markedly above average hearing). The onus is on you.
 
Before 16/44 and 24 /96 we were blessed with analog. What some have defined as true audio(without the needle or tape noise). 16Bit sounds are going to be less defined than 24Bit, if not via your ears than via math as 24bit support more harmonics and tones. Although 44khz sounds fine 48khz defines better siblence and highs. I find the 96/khz range really only revellent in mixing (coloring) multi-tracks of audio.As far as 192khz. I think a 30/32Bit@48/96Khz would be better BUT until we are capable of capturing a true sound wave digitally, we will always only have BITS of it. Speeding it up from 44 to 96 or even 196khz may improve the ability to hear more slight frequencies but it will always be a matter of how many BITS of info are available.
 
...16Bit sounds are going to be less defined than 24Bit, if not via your ears than via math as 24bit support more harmonics and tones...

Bit depth is about dynamic range and not harmonics or tones. 16 bit is 96dB and 24 bit is 144db. Each bit is 6dB of dynamic range. Using too few bits can add quantization errors though.
 
but personally I'd like to see some 192khz music, and maybe even expand to a 32bit bit depth, but the holy grail would be 32bit DXD (aka 352.8Khz)

Real 32 bit is simply not possible, no hardware controller can deal with true 32 bit audio at high bitrate.
Even your computer , when you open en 32bit float, it wil ldump the last 8 bits and play 24 bits but sure not the full 32bit float.

Float means, and that's where it's good for, in case of mixing audio and adding effects of any sort, It can improve the 24 bit result cause part of the rubbish that would be there at real 32bit can be discarded.

DXD is still 24 bit but okay it can handle also 32 bit float but will rip the last 8 bits also.

I really doubt that 32 bit sound chips will come cause, lets be frankly, 24bit at 192kHz is still anno 2017 not widestream available.

Do they try? Sure but whatever the test, the digital noise is the biggest problem.
Even DXD with their 340kHps inject a higher digital noise in the audio file with as results a noise floor of unacceptable -70dB

And last, High sample rates are useless if the audio file is recorded ad far lower kHz.
Second problem, 24bit/192kHz are very big files so even then there is the risk that they compress the file
so back to more distortion.

24bit /192kHz is possible but only for very expensive equipment , huge storage spaces, and high quality forced cooled A/D convertors. Yes these can get quite hot.

Oh well, 24bit 96khz will be here still several years before the lossless plunge of 192kHz is taken.
[doublepost=1505641787][/doublepost]
Bit depth is about dynamic range and not harmonics or tones. 16 bit is 96dB and 24 bit is 144db. Each bit is 6dB of dynamic range. Using too few bits can add quantization errors though.

Theoretical yes but practical?
Look at the real noise floor:
CD usual around -91db, 24 dB around -110 db
Only very expensive equipement reaches -130dB but sure no -144dB
[doublepost=1505642505][/doublepost]
Before 16/44 and 24 /96 we were blessed with analog. What some have defined as true audio(without the needle or tape noise). 16Bit sounds are going to be less defined than 24Bit, if not via your ears than via math as 24bit support more harmonics and tones. Although 44khz sounds fine 48khz defines better siblence and highs. I find the 96/khz range really only revellent in mixing (coloring) multi-tracks of audio.As far as 192khz. I think a 30/32Bit@48/96Khz would be better BUT until we are capable of capturing a true sound wave digitally, we will always only have BITS of it. Speeding it up from 44 to 96 or even 196khz may improve the ability to hear more slight frequencies but it will always be a matter of how many BITS of info are available.

Analog okay, we had no choice but using Tape drives and LP's.
Their noise floor was far from great but their harmonics a blessing and analog to analog is no conversion.
CD: the problem is the 44.1kHz sample rate , some people with good hearing find them annoying.
The search for higher sample rates is going the wrong direction.
What happens at 96kHz and 192kHz, the master file can go instead of 20kHz to 30/40/50... kHz sonic range.
A dog or cat will sure not like this.
[doublepost=1505643595][/doublepost]
Why waste time reading a BS article when it's obvious to anyone with a pair of functioning ears that 24/96 is better?

Well, if the original recording was made in 24bit/96 kHz then sure this will be better the CD audio.
Even when digital distortion becomes a first concern and good audio card can filter out this digital noise
but the card must support then 32bit float.

When we humans hear the most disortion ? At low volumes.
Now compare an original recording, then a remastering and again a remastered file.
They get louder and louder.

One negative point of digital, clipping above 0db (commercial devices)
Clipping means, pure DC is going to the loudspeakers , ok a fraction of a second but at loud volumes .....

in 2017 there is still too many CD mastered 16bit/44.1kHz up sampled on sale.
Therefore, I sure would not say that 24bit/96kHz is de-fact better cause they will never tell you in which format the original recording was.

Oh and, if the re-sample the analog source with 24bit/96kHz will this be better then the analog source?
Sure not.
 
Real 32 bit is simply not possible, no hardware controller can deal with true 32 bit audio at high bitrate.
Even your computer , when you open en 32bit float, it wil ldump the last 8 bits and play 24 bits but sure not the full 32bit float.

This is certainly true. There are and never will be any playback equipment that can handle the range of a 32-bit floating audio file. Just think what would happen if a playback system could play back 1/10th of the dynamic range possible in a 32-bit float file. Such a system could be used as a weapon.

The dynamic range of a "float" is 640 db Compare this with a typical atomic bomb which is in the range of 200 to 278 db. 640 is closer to the Star Wars "Death Star" than an atomic bomb. 2^38 is a really big number

So why one Earth do we have 32-bit float files? Because we process audio data internally in this format. We do it so that internal, intermediate values don't clip. We all know that x = arctan(tan(x)) but computers don't do algebra and would calculate the function as written and we can see some values of x would clip. Yes silly computer math.

Andyes there are audio processing functions that use trig functions, recursive filters and even Fourier transforms.

32 bit integers actually have more resolution and a lower noise floor then 32 bit floats but we used the floats for the high dynamic range. It is a USLESS distribution format but good for processing and use in a studio because it records exactly what is inside the computer when it is processing the audio, so we can save intermediate results with zero los..

But it is obvious that 32 bits of dynamic range will never is reproducible.

A distribution format only has to record what is reproducible. Some would argue it only has to record what is within the range of human hearing. 24-bit 96k goes well past both criteria.
 
32bit integers are a matter of software.
But as soon when you click on play then the audio card receives maximum 24 bit.

I record audio on 32 bit float format which is 24bit audio + 8 zero's at the end.
The noise floor is at -123db , hardware can never reach theoretical calculations.

But as soon effect or reverb or whatever is added, then these last 8 bit contain data from the calculated audio.
And the result is then once the track is mastered then the the 24 bit recording is a little bit better then recording at 24 bit and calculations there.

But working with 32 bit float asks faster computers and very good sound controllers.
True, clipping on pro audio is not on 0db but as soon is remaster for consumer devices keep always below 0dB due to less expensive audio controllers and Injecting DC (clipping) to speakers is disastrous. Better amps will switch off if the detect DC.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.