[skamp]

But I am also inclined to defer to the wisdom of the engineers who did extensive research when they designed the Blu-ray spec. They seemed to think it can make a difference otherwise they would not have bothered to include support for 192k in their spec.

Did it never occur to you that they only went for higher numbers for marketing reasons? Admitting that 16/48 was already good enough would have been detrimental to sales. They had to make people believe that higher numbers really were that much better.

As far as 24 bits vs 16 bits, it is pointless, nay it is impossible to discuss this as long as people insist that more bits can ONLY mean LOUDER bits rather than being set to produce an equal volume range at a finer gradation

Quieter, actually, since the point of reference is always 0 dBFS, which is equally loud at 24 and 16 bits.

 

[skamp]

it is impossible to discuss this as long as people insist that more bits can ONLY mean LOUDER bits rather than being set to produce an equal volume range at a finer gradation

Here's this "finer gradation" you're talking about.

Let's take the loudest signal possible, which is 0 dBFS. It is encoded as a signed 16 bit integer and a signed 24 bit integer like this:

16 bit: 01111111 11111111
24 bit: 01111111 11111111 11111111

The 16 bit value is missing all the values in between that can be encoded thanks to the lower 8 bits in the 24 bit format. So the range of values that's missing is between the following two numbers:

01111111 11111111 11111111
01111111 11111111 00000000

Now, how much difference is that? What is the range of "finesse", "nuance", "precision" that is gained with 24 bit sampling? First, we have to convert those binary numbers to decimal. Then, the formula to get an equivalent value in dBFS is this:

20 * log10(n / (2^23 - 1))

Here are the converted values:

01111111 11111111 11111111 = 0 dBFS
01111111 11111111 00000000 = -0.0003 dBFS

That's a difference of 0.0003 dB! It's very, very, very small, well beyond audibility. Now, let's see how much range we're gaining at various levels:

01111111 11111111 11111111: 0 dBFS,
01111111 11111111 00000000: -0.0003 dBFS (0.0003 dB more range)

00111111 11111111 11111111: -6.0206 dBFS,
00111111 11111111 00000000: -6.0211 dBFS (0.0005 dB more range)

00011111 11111111 11111111: -12.0412 dBFS,
00011111 11111111 00000000: -12.0423 dBFS (0.0011 dB more range)

00001111 11111111 11111111: -18.0618 dBFS,
00001111 11111111 00000000: -18.0639 dBFS (0.0021 dB more range)

00000111 11111111 11111111: -24.0824 dBFS,
00000111 11111111 00000000: -24.0866 dBFS (0.0042 dB more range)

00000011 11111111 11111111: -30.103 dBFS,
00000011 11111111 00000000: -30.1115 dBFS (0.0085 dB more range)

00000001 11111111 11111111: -36.1237 dBFS,
00000001 11111111 00000000: -36.1406 dBFS (0.0169 dB more range)

00000000 11111111 11111111: -42.1443 dBFS,
00000000 11111111 00000000: -42.1782 dBFS (0.0339 dB more range)

00000000 01111111 11111111: -48.1651 dBFS,
00000000 01111111 00000000: -48.2329 dBFS (0.0679 dB more range)

00000000 00111111 11111111: -54.1859 dBFS,
00000000 00111111 00000000: -54.3222 dBFS (0.1363 dB more range)

00000000 00011111 11111111: -60.2071 dBFS,
00000000 00011111 00000000: -60.4818 dBFS (0.2747 dB more range)

00000000 00001111 11111111: -66.2287 dBFS,
00000000 00001111 00000000: -66.7872 dBFS (0.5585 dB more range)

00000000 00000111 11111111: -72.2514 dBFS,
00000000 00000111 00000000: -73.407 dBFS (1.1556 dB more range)

00000000 00000011 11111111: -78.2763 dBFS,
00000000 00000011 00000000: -80.7666 dBFS (2.4903 dB more range)

00000000 00000001 11111111: -84.3054 dBFS,
00000000 00000001 00000000: -90.309 dBFS (6.0036 dB more range)

00000000 00000000 11111111: -90.343 dBFS,
00000000 00000000 00000001: -138.4738 dBFS (48.1308 dB more range)

As you can see, the ranges are extremely small until the signal gets quieter and quieter, most (if not all) way beyond audibility.

 

[bigshot]

Blurays often have as many as 8 channels. They need a higher bitrate. The rest is marketing. Home theater folks can be as nuts about numbers as audiophiles.

 

[dbbloke]

Quote:

No, you need a sampling frequency greater than two times the maximum frequency to be able to reconstruct the original signal. The reconstruction is perfect in theory. It seems you don't understand the theorem.
 
 
No, that's wrong. See reconstruction.
 
 
No, it's not. You don't understand the sampling theorem.
 
 
It seems you're talking about ancient non-oversampling DACs.
 
 
All DACs (ignoring the ancient nos crap) I know of perform worse at 176.4 or 192 kHz sampling rate.
 
 
LOL, the irony! You don't get increased resolution. Please read up on the sampling theorem and oversampling DACs. Also see #846.

 
Quote:

That's why I was specifically talking about DAC chips.
 
It is impossible to discuss this with people that do not understand the sampling theorem, quantization and dithering. Please read the first post.
 
You're talking about "equal volume range". What's that range? What's the math behind the number you come up with?
 
Hint: more bits = higher dynamic range
 
Set a wallpaper with smooth gradients on your desktop and switch from 32 bit to 16 bit colors. Do you really not see the difference?
 
Afaik nobody argued that.

It's impossible to discuss with people who don't listen to live acoustic music often, who haven't heard a well recorded 24/96 on a really good 2ch speaker hifi system but base their assumptions in science. :>
 
And to note, probably the simplest and best way to do this is to take a live instrument (like a harpsichord/harp imho), record 16-44/48 then 24-96/192 bit. Play both files back and see if the instrument sounds different in any way.
Does one sound more detailed than the other, does the 3d staging and room dynamics sound different, does the instrument have more air about it, is it more solidly placed in the room, does one seem wider and larger, can you hear the reactions between the strings or the fingers plucking sound more finger like as it would in real life.
Assuming same recording and playback hardware (preferably underground in a faraday cage ) forget the track is different, the instrument sounds slightly more alive and real. At least this is what I heard, blind, back to back, levels normalised. Every time it was easy to write down what I thought and compare after doing said test a dozen or more times and find I was correct. Lucky biased guessing? I guess 1 in 4000 for a dozen comparisons is possible. What's the science on probability say? Hey, some science is mostly right, just some is more flawed then others, not all as I clearly say. I actually dropped a coin and it landed on it's side just the other day. Think it was a zlotie though.
 
Using some conversion software pulls so many things into the equation it makes the argument open to criticism.

 

[chewy4]

So using conversion software pulls too many things into the equation but... using two different recordings doesn't?

 

[xnor]

Quote:

It's impossible to discuss with people who don't listen to live acoustic music often, who haven't heard a well recorded 24/96 on a really good 2ch speaker hifi system but base their assumptions in science. :>

You cannot know that, but regardless, it doesn't change a bit that you make claims/assumptions/statements (or whatever you like to call it) about something you don't understand and say others got it wrong.
 
 
Quote:

And to note, probably the simplest and best way to do this is to take a live instrument (like a harpsichord/harp imho), record 16-44/48 then 24-96/192 bit.

The problem with this has been pointed out before, but you added another variable: different sampling frequencies.
 
Btw, tests have been done before ...
 
 
Quote:

Assuming same recording and playback hardware (preferably underground in a faraday cage ) forget the track is different, the instrument sounds slightly more alive and real. At least this is what I heard, blind, back to back, levels normalised. Every time it was easy to write down what I thought and compare after doing said test a dozen or more times and find I was correct. Lucky biased guessing?

Please share the recordings used for the test and the ABX logs. Then we can try to figure out what really happened and if it was lucky guessing, some error elsewhere, or not.
 
 
Quote:

I guess 1 in 4000 for a dozen comparisons is possible. What's the science on probability say?

How many comparisons did you do? How many did you get right? Again, please share the ABX log if possible.
 
 
Quote:

Using some conversion software pulls so many things into the equation it makes the argument open to criticism.

No, just no. /wallbash

 

[skamp]

 

[bigshot]

It's impossible to discuss with people who don't listen to live acoustic music often, who haven't heard a well recorded 24/96 on a really good 2ch speaker hifi system but base their assumptions in science. :>

I listen to classical music most of the time, and I've not only heard 24/96, I've supervised recording sessions and mixes that were produced in 24/96. The difference between 24 and 16 is dynamic range, not resolution.

 

[AstralStorm]

The main limitation is the microphone equipment, those have ~85 dB SNR with 1 Pa = ~95 dB. The best ones achieve 90 dB SNR.
Note that orchestra can almost reach 80 dB SNR in a full attack.
The difference between 16-bit and 20-bit summed to signal (AWGN) won't move the dynamic range even by 1 dB.

 

[XxDobermanxX]

 

[bigshot]

The main limitation is the microphone equipment, those have ~85 dB SNR with 1 Pa = ~95 dB. The best ones achieve 90 dB SNR.

Some stuff plugs in too. It's helpful to be able to boost a channel's level in the mix without pulling up noise along with it. The more dynamic range the merrier,

 

[AstralStorm]

Indeed, but that has nothing to do with bit depth, since amplification is typically analog.

 

[bigshot]

No, you usually normalize tracks up in the digital domain.

 

[kiteki]

/
I accept that 16/44.1 CD format is adequate, in theory, for almost everybody. I'm also willing to believe that some people perceive (not HEAR, but PERCEIVE) frequencies above 20kHz.

 
According to one study, people can hear up to 25kHz at extreme volumes.  At normal volumes, one theory is we can perceive UHF via some other pathway, like eyes or skin, yes, which means that it wouldn't work with headphones.
 
The original Oohashi experiment did note it didn't work with headphones, and the study has never really been perfectly replicated and refuted... that is, with fMRI, gamelan music, super-tweeters, whateverz... so the findings still stand in that sense.
 
I suppose there are some perceptions which are too complicated to ask "yes / no" and tick boxes, you need to look inside their head hahaha.  It's like subliminal advertising... "did you just see a subliminal advertisement? Yes / No" ...............................
 
 
Perhaps... if a room is painted blue, versus red, and you're trying to find out if the colour of a room affects peoples mood, without telling them what the study is about, you usher 20 people into different coloured rooms and make them tick boxes, you see that's difficult as well, since it's a study on emotion there are no clear-cut limits, unless you want to measure levels of cortisol in their blood after being in a red room for hours.
 
 
Anyway there is one modern paper which 'proves' that we can perceive higher-rez material than 44.1 kHz, but the statistics aren't very good in it, I'm not sure if it's valid, in my opinion (if it's invalid, I suppose that says something about the validity of all other papers listed at the AES) --> http://www.aes.org/e-lib/browse.cfm?elib=15398
 
What I did like about it, is they didn't use the common fast switch, time-aligned ABX, which imho is very overrated.  Someone really needs to write a new blind-test program for Foobar, without the immediate switch (like a 10 second break between A... A... A, versus A... B... B..., and not time-aligned, the sample is played from 0 - 20s, etc.).

 

[bigshot]

Quote:

 
According to one study, people can hear up to 25kHz at extreme volumes.

 
As pain.