or Connect
Head-Fi.org › Forums › Equipment Forums › Sound Science › 24bit vs 16bit, the myth exploded!
New Posts  All Forums:Forum Nav:

24bit vs 16bit, the myth exploded! - Page 149

post #2221 of 3422

I think most Pentatone releases are recorded DSD.

post #2222 of 3422
Quote:
Originally Posted by bigshot View Post
 

I think most Pentatone releases are recorded DSD.


I think the Pentatone Remaster Classics referred to above are remastered to DSD from the original analog tapes. The founders purchased the rights and the tapes from Philips, which did the original recordings.

post #2223 of 3422

Yes, but everything Pentatone does themselves is DSD based. (not that it makes a lick of difference.)

post #2224 of 3422
In my "audiophile" experience, the only thing that really made a sonic difference is the mastering of the music.

I've come to realize that I am at the mercy of the sound engineers/producers for the actual quality of the music that reaches my ears.
post #2225 of 3422

I totally agreed, but it also depends on the original mix master as well, if it was very badly produced or damaged, nothing in this world could make it better.

 

Quote:

Originally Posted by Lespectraal View Post

In my "audiophile" experience, the only thing that really made a sonic difference is the mastering of the music.

I've come to realize that I am at the mercy of the sound engineers/producers for the actual quality of the music that reaches my ears.
post #2226 of 3422
Quote:
Originally Posted by Danz03 View Post

I meant to include that as well.
post #2227 of 3422

Great thread, not at all surprised to hear that the folks who made the CD standard knew what they were doing. Of course, this requires some care in the ADC and post-processing that are the final steps in digital audio mastering.

 

I know it is a somewhat different topic...but curious to hear more thoughts: how does bit-depth affect the linearity of a digital sound reproduction system? Of course, the sound wave (speaker to ear) is itself linear, but what happens between the digital file and the movement of the speaker cone is not necessarily linear.

post #2228 of 3422
Quote:
Originally Posted by JHern View Post

Great thread, not at all surprised to hear that the folks who made the CD standard knew what they were doing. Of course, this requires some care in the ADC and post-processing that are the final steps in digital audio mastering.

I know it is a somewhat different topic...but curious to hear more thoughts: how does bit-depth affect the linearity of a digital sound reproduction system? Of course, the sound wave (speaker to ear) is itself linear, but what happens between the digital file and the movement of the speaker cone is not necessarily linear.
http://youtu.be/nLEhfieoMq8
If I understand your question correctly then this video answers it.
post #2229 of 3422
Quote:
Originally Posted by James-uk View Post


http://youtu.be/nLEhfieoMq8
If I understand your question correctly then this video answers it.


That was a great video; more clear and concise than other examples I've seen and is one I will share when asked about 24-bit audio.

post #2230 of 3422
Quote:
Originally Posted by Krutsch View Post
 
Quote:
Originally Posted by James-uk View Post


http://youtu.be/nLEhfieoMq8
If I understand your question correctly then this video answers it.


That was a great video; more clear and concise than other examples I've seen and is one I will share when asked about 24-bit audio.


the video is ok and the conclusions are obviously right. but it takes some shortcuts. a few people already find that Monty was taking too many shortcuts in his vids, so they would certainly not be happy about this one.

also there are a few stuff I'm not sure I agree with. the most obvious is when he uses a DR value from the DR database, and takes it as the actual dynamic range of that track at the end. it's easy to find songs that have about 50db of dynamic range, but they won't get 50 as DR database value because they don't measure the same thing. for the video it's ok because the actual dynamic of those tracks are still well under 16bit, so the conclusion still works just fine.

post #2231 of 3422
Quote:
Originally Posted by James-uk View Post


http://youtu.be/nLEhfieoMq8
If I understand your question correctly then this video answers it.


Yes, nice simple video that clears up the basics very well!

 

Avoid the comments though, trust me. I wonder if we know this "ezrazski" guy? His arguments sound so familiar.

post #2232 of 3422
Quote:
Originally Posted by castleofargh View Post
 


the video is ok and the conclusions are obviously right. but it takes some shortcuts. a few people already find that Monty was taking too many shortcuts in his vids, so they would certainly not be happy about this one.

also there are a few stuff I'm not sure I agree with. the most obvious is when he uses a DR value from the DR database, and takes it as the actual dynamic range of that track at the end. it's easy to find songs that have about 50db of dynamic range, but they won't get 50 as DR database value because they don't measure the same thing. for the video it's ok because the actual dynamic of those tracks are still well under 16bit, so the conclusion still works just fine.


Sure... but that is getting hung-up on details that are meaningless to a target consumer audience (e.g. my dad or a non-audiophool friend). Believe me, I am getting an increasing number of questions/comments from people about audio stuff that never before cared. I was in a work meeting last Friday with a storage engineer that I really respect and, out of nowhere, he starts telling me about his new John Coltrane SACD and how great it sounds (but he's puzzled about the whole copy protection / down-sampling thing with SACD and digital out).

 

For someone that wants to argue with you about DR values, and starts with a knowledge base and a set of expectations, send them right to the AES convention papers of your choosing.

 

IMO, what makes this video compelling is the lack of a condescending tone (e.g. like just about everything else I've read/seen). I really appreciated the illustrated "lollipop" diagram (which quickly covers the folly of stair-stepping, that we all learned in calculus, as an abstract representation of audio resolution), along with the highly demonstrative examples of quantization noise. That latter piece was nicely done and really drives home the point that those extra bits are just reducing noise.

post #2233 of 3422
Quote:
Originally Posted by James-uk View Post


http://youtu.be/nLEhfieoMq8
If I understand your question correctly then this video answers it.


A nice video, but no, it doesn't explain my question.

 

The 16-bit vs 24-bit discussion thus far focuses entirely on the recording and digital mastering product. I completely agree with the main argument that 16-bit is more than sufficient as the mastering product. I have an advanced mathematical education so I have no issues understanding this part.

 

My question has to do with the practical effect on electronics going in the opposite direction, what happens from the digital audio file to the speaker cone, and whether using a higher bit depth (perhaps in combination with variable/adaptive gain) at any point in this process is useless, or not? For example...my favorite-sounding audio player, Audirvana, uses 64-bit (variable gain) for internal processing (they say it is to avoid round-off errors)...is this just a waste of my computer's CPU time? Another example, is buying anything more than a 16-bit depth DAC also a complete waste of money? Could it be that all the digital processing in this stage is just nonsense, that the sound improvements we hear in players like Audirvana are just a matter of coloring/styling the output to sound more pleasing to the ear? Should we focus more attention on the pre-amp stages in the DAC chip rather than worrying about all of the digital machinations of the chip?

 

I hope everyone can see that these are different questions. But I fear that the distinction might be lost on some, and if there is value to the A=>D=>A process (at any point) in using higher bit-depths then let's be careful to not throw the baby out with the bathwater.


Edited by JHern - 1/25/15 at 5:32pm
post #2234 of 3422
Quote:
Originally Posted by JHern View Post
 

Audirvana, uses 64-bit (variable gain) for internal processing (they say it is to avoid round-off errors)...is this just a waste of my computer's CPU time? Another example, is buying anything more than a 16-bit depth DAC also a complete waste of money?

If you are passing the data through multiple filters, then each stage will add up to half a LSB of error. So working in a larger bit space will allow you to apply filters without compounding the noise & distortion. Plus, some DSP algorithms require far more than 24 bits for intermediate computational products just to get final results accurate to near 16 or 24 bits.

 

Why you would need 64-bit I have no idea, but I don't expect there is any penalty if you are using a 64-bit processor anyway (as most of us are these days). The data will be down sampled to 16 or 24 bit in the end, and will have up to half an LSB of error in the end, so if your DAC can handle 24 bit then better to output your data as 24 bit, though you probably won't hear the difference anyway.

 

Good question, btw!


Edited by lamode - 1/25/15 at 6:28pm
post #2235 of 3422

DSP / Digital Filter algorithms use a basic element called a multiply-accumulate (MAC).  The idea is to multiply two numbers then store the result and add it to the next result (accumulate).  When you multiply two 16-bit numbers, you need 32 bits to store the full result if the values are large enough. Each time you add, in theory if the values are large enough you can need an extra bit for accumulate (since adding the two largest values is like doubling which is 1 bit in binary arithmetic).  For this reason standard DSPs feature a 16x16 into 40 bit MAC.  For audio you can do 24x24 into 48 or 60 bits, or in your case 64.  Often times they round up the accumulate word length since it is easier for computers to work on powers of 2 like 32 and 64 than other word lengths.

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › 24bit vs 16bit, the myth exploded!