Trouble is, that's exactly what we're not getting ... I haven't seen a single piece of writing yet that has explained what should be measured, to get a good correlation with what subjectively registers ...
I don´t think this is correct. Yes you can hear signals below the noise floor IF they are there. I don´t think that a mic can actually record below the noise floor, seems a contradiction.
Jan
Attachments
This concept of noise floor can be inherently misleading. The 20dB microphone noise floor is actually total integrated noise 20-20k, any single tone has a gain of ~43dB over that (in a 1Hz BW). So in fact you can fiddle with FFT length and show any answer you want and I have called out folks doing just that.
Neither have I. so i'm going to explore this stuff myself and see what I find which might make a difference. I'm certainly not going to devote my whole life to it, though. I'm just going to do a simple look at everything in the output and its total level and see what number that comes up with and if significant.
Noise floor..... electronic signal noise voltage or the SPL 0 db ?
THx-RNMarsh
Last edited:
The difference between Toslink and a high speed connection is extensive. Toslink uses an LED and an IR sensing transistor similar to a photo interrupter. Telcom links use laser diodes with modulators and special receivers. The LED cannot switch faster than the data rate of a 96K SPDIF stream unless its selected. Literally no data comes out at 192K on most Toslink transmitters. The problem is that the LED does not turn on and off symmetrically (its actually always on but modulated).
However current SPDIF receivers can get jitter below 20 pS pretty consistently even with Toslink. That 20 pS number is considered inaudible in all the tests of jitter audibility.
SOTA DAC chips can deliver the same distortion at 44.1 to 192. They measure differently because the measurement bandwidth changes. On the good chips avoid the top few dB and the distortion is well below any accepted standard of audibility.
However I'm sure many on this thread will easily be able to hear the jitter and distortion even though the energy will be far below what is necessary to move a physical diaphragm.
The question is then what is the discriminating ability of our aural perception.
I am afraid that such an ‘equivalent FFT length’ varies among individuals, over time and according to the mood for each individual
No emoticon Demian?
George
You can help it on its way though with a modified LED drive stage. I built this many years ago using nothing more than a high brightness miniature LED and TOSLINK coupler (to make my own transmitter)
http://www.diyaudio.com/forums/digi...ptical-diy-transmitter-super-clean-drive.html
At the end the usual blather about how bit perfect copies can sound different. Yawn, I don't care who says it it's nonsense.
That does not nullify what he said, as you might want it to, esp regarding 24b is better than 16b.
Never-the-less, every mastering person I have read or contacted says the same thing..... 24b systems sound better. Katz can at least hear differences very well and produces some of the best recordings so it does matter what he hears.
BTW - That comment was not meant to be taken as a literal engineering statement, Scott. Rather, it has become a phrase - almost a joke - to mean everything is Not perfect in the digital world - at least when it comes to bits and which sounds better or more perfect. Yes, its a cheap shot, but he hears the imperfections.
THx-RNMarsh
Last edited:
That article is from 1998! Back then HDCD still had a chance to deliver 24bit word lengths from std CDs yet he makes no mention of it. At the time he clearly thought that DVD-audio would take over. That didn't happen!
Telling for me it "[FONT=Arial, Helvetica]In fact, my experiments and those of others firmly point out that the sonic improvements we hear at the higher sampling rates are probably not due to the increased bandwidth (i.e., hearing frequencies that we previously could not hear), but rather to the reduction in artifacts in the 20-20 kHz band because the digital and analog filters don't have to work as hard." Once again completely forgetting about oversampling in the DAC. A very 1985 argument.
[/FONT]
Telling for me it "[FONT=Arial, Helvetica]In fact, my experiments and those of others firmly point out that the sonic improvements we hear at the higher sampling rates are probably not due to the increased bandwidth (i.e., hearing frequencies that we previously could not hear), but rather to the reduction in artifacts in the 20-20 kHz band because the digital and analog filters don't have to work as hard." Once again completely forgetting about oversampling in the DAC. A very 1985 argument.
[/FONT]
It doesnt matter what his marketing predictions or hopes were
.
Artifacts are artifacts and they seem to be identified one by one and reduced over time. Things are better in many ways.... in all of audio. Who knows what causes the difference in sound of 16 and 24b systems? There are no audible differences, is one answer. But, there are apparently still artifacts. And, all the mastering people think 24b systems sound better. They seem to for me as well. Why do they hear it and identify it or tag it with 24b systems as better? Better yet.... what goes along with 24b systems that could cause fewer or lower levels of artifact(s)?
Take a look at the BenchMark ADC1 USB. It can produce several output bit levels and sampling rates simultaneously. Great for mastering houses but that feature could be interesting for evaluating 16/24b.
THx-RNMarsh
.
Artifacts are artifacts and they seem to be identified one by one and reduced over time. Things are better in many ways.... in all of audio. Who knows what causes the difference in sound of 16 and 24b systems? There are no audible differences, is one answer. But, there are apparently still artifacts. And, all the mastering people think 24b systems sound better. They seem to for me as well. Why do they hear it and identify it or tag it with 24b systems as better? Better yet.... what goes along with 24b systems that could cause fewer or lower levels of artifact(s)?
Take a look at the BenchMark ADC1 USB. It can produce several output bit levels and sampling rates simultaneously. Great for mastering houses but that feature could be interesting for evaluating 16/24b.
THx-RNMarsh
Last edited:
OK. No difference. And, they are selling something.
I only did the downloads for convenience to have the music accessible anywhere I travel as well as listen at home.
I have only been digging up info and mastering lab views etc since I heard such a nice improvement on my home system.
So, I have to give these two ideas a thumbs down.
It is clear now that there are no answers to be found here at this time.
The closest to hardware apples to apples comparison of bits might be the ADC 1 USB. If I dont hear any difference with it between bits then I will conclude that everything gets tightened up in the design of the next gen product, such as 24b resulting in fewer or reduced artifacts. Any differences heard with typical CD players vs direct 24b files are in the artifacts of that generation product. And, I would not expect the $200-300 typical CD player to use all available means and parts to reduce all the artifacts to the level 3-4K dollar worth of converters would use. IMBG.
THx-RNMarsh
I only did the downloads for convenience to have the music accessible anywhere I travel as well as listen at home.
I have only been digging up info and mastering lab views etc since I heard such a nice improvement on my home system.
So, I have to give these two ideas a thumbs down.
It is clear now that there are no answers to be found here at this time.
The closest to hardware apples to apples comparison of bits might be the ADC 1 USB. If I dont hear any difference with it between bits then I will conclude that everything gets tightened up in the design of the next gen product, such as 24b resulting in fewer or reduced artifacts. Any differences heard with typical CD players vs direct 24b files are in the artifacts of that generation product. And, I would not expect the $200-300 typical CD player to use all available means and parts to reduce all the artifacts to the level 3-4K dollar worth of converters would use. IMBG.
THx-RNMarsh
Last edited:
Little by little…
George
Thanks George, I suspect you understood my point.
Scott,
Using an analyzer you can get 1 hertz bandwidth. For our ears we use what are called critical bandwidths, which do vary with frequency.
Once again, incorrectly conflating the difference between recording (where 24 bits is very useful) and data delivery to the listener (where 16 bits is more than sufficient unless you're doing something pathological). At this point, I can't help but think that it's not confusion, but deliberate obfuscation, much in the manner of claiming that SPL threshold curves taken under conditions of quiet are applicable in the presence of signal.
- Status
- Not open for further replies.