Oversampling vs. Non-Oversampling DAC
- Thread starter Netdewt
- Start date
very controversial subject. I like to up sample typically all the way to 24/192 from 16/44. It typically brings out some depth and width in the soundstage and instruments tend to separate out from each other. There may be some loss in solidity of some instruments. IMHO it may not add quality in some cases but does not cause much harm if any that I notice in my system #2.
I think some DACs sound better at up sampling than others. I like what my music hall does...
I think some DACs sound better at up sampling than others. I like what my music hall does...
Right. From what I've read, the NOS camp says that oversampling ruins the sound because it puts information in that is not there. Some of them say that NOS DACs sound more like vinyl.
As someone who makes a living working with digital photography, I have to "oversample" images from time to time (we call it res-ing up). In theory, I would tend to lean towards what you said: oversampling is better if it's done right, but I have not had the opportunity to audition multiple DACs side by side.
I'm excited to hear what more AKers think. You guys tend to be more down to earth than many of the other forums.
As someone who makes a living working with digital photography, I have to "oversample" images from time to time (we call it res-ing up). In theory, I would tend to lean towards what you said: oversampling is better if it's done right, but I have not had the opportunity to audition multiple DACs side by side.
I'm excited to hear what more AKers think. You guys tend to be more down to earth than many of the other forums.
riverrat
professional camper
You've already demonstrated the ability to search the web, so you probably checked out head.fi already. That gang probably has a lot of opinions on the matter, although I have not researched that particular question yet.
I've been leaning toward the Music Hall that stoutblock has, but by the time I make my mind up there will probably be something better and less expensive, e.g. the tube Jolida unit allegedly coming this spring..
I like AK a lot too, seems to be a lot of folks here that share my take on things. I would venture to say that the degree to which we've jumped into PC-based playback is probably less overall than on some forums, simply because we are into vintage (mostly) and older (mostly).
I've been leaning toward the Music Hall that stoutblock has, but by the time I make my mind up there will probably be something better and less expensive, e.g. the tube Jolida unit allegedly coming this spring..
I like AK a lot too, seems to be a lot of folks here that share my take on things. I would venture to say that the degree to which we've jumped into PC-based playback is probably less overall than on some forums, simply because we are into vintage (mostly) and older (mostly).
crackerkorean
Striving for Polymathdom!
Ok here is some on the technical side of things.
There is a difference between oversampling and upsampling. You cant oversample a CD since the number of samples per second are finite. The time between two samples on any redbook CD is exactly the same as any other. What CD players do is interpolate between two samples filling in the missing information.
The interpolation between the two samples can happen different ways (linear, just the repeating of a single data bit, curve fit etc). I think most of these oversampling dacs use just repeat the data. I am almost 100% sure the TDA1541 does.
In response to the putting information thats not there comment, if there is a peak in-between two samples the frequency content of that peak is beyond human hearing so is considered negligible data.
The advantage to oversampling DACs is the fact that you extend the frequency range that the low pass filtering happens. This is easier and cheaper to do.
Non OS = the filter has to be between 20k and 22.1k (2.1k bandwidth is very narrow)
8x OS = will increase that bandwidth by 8 (16.8k)
I some implementations just take an OS dac and make it NOS adding all kinds of high frequency content that isnt supposed to be there but some say it sounds better. Its not correct on the side of technical but if they like it, thats fine.
As I have said before I think most of the differences in sound come from the analog stage on the output side of things as opposed to digital side.
If you want to be playing back correct data just make sure the filter is cutting off the information properly.
There is a difference between oversampling and upsampling. You cant oversample a CD since the number of samples per second are finite. The time between two samples on any redbook CD is exactly the same as any other. What CD players do is interpolate between two samples filling in the missing information.
The interpolation between the two samples can happen different ways (linear, just the repeating of a single data bit, curve fit etc). I think most of these oversampling dacs use just repeat the data. I am almost 100% sure the TDA1541 does.
In response to the putting information thats not there comment, if there is a peak in-between two samples the frequency content of that peak is beyond human hearing so is considered negligible data.
The advantage to oversampling DACs is the fact that you extend the frequency range that the low pass filtering happens. This is easier and cheaper to do.
Non OS = the filter has to be between 20k and 22.1k (2.1k bandwidth is very narrow)
8x OS = will increase that bandwidth by 8 (16.8k)
I some implementations just take an OS dac and make it NOS adding all kinds of high frequency content that isnt supposed to be there but some say it sounds better. Its not correct on the side of technical but if they like it, thats fine.
As I have said before I think most of the differences in sound come from the analog stage on the output side of things as opposed to digital side.
If you want to be playing back correct data just make sure the filter is cutting off the information properly.
A few threads I found:
http://www.computeraudiophile.com/content/Oversampling-DAC-vs-NON-Oversampling-DAC
http://www.head-fi.org/forums/f7/over-sampling-vs-non-over-sampling-dac-w-computer-source-383467/
http://www.tnt-audio.com/sorgenti/audionote11kit_e.html
This article is very convincing to me. I am definitely leaning toward finding something NOS, although I'm sure I could find something satisfying either way.
http://www.sakurasystems.com/articles/Kusunoki.html
It seems to explain in detail what crackerkorean just posted.
"Originally, oversampling was developed to allow the use of an analog filter with gentler characteristics as a post-filter, and not to increase the amount of information. Many people still misunderstand this."
http://www.computeraudiophile.com/content/Oversampling-DAC-vs-NON-Oversampling-DAC
http://www.head-fi.org/forums/f7/over-sampling-vs-non-over-sampling-dac-w-computer-source-383467/
http://www.tnt-audio.com/sorgenti/audionote11kit_e.html
This article is very convincing to me. I am definitely leaning toward finding something NOS, although I'm sure I could find something satisfying either way.
http://www.sakurasystems.com/articles/Kusunoki.html
It seems to explain in detail what crackerkorean just posted.
"Originally, oversampling was developed to allow the use of an analog filter with gentler characteristics as a post-filter, and not to increase the amount of information. Many people still misunderstand this."
Last edited:
Its all over my head. All I know was the Music Hall is selectable from 16/44, 24/96 and 24/192. 24/96 is a noticeable improvement basically always and 24/192 maybe less noticeable. I have a few test tracks that tend to have a bigger impact than others.
the-real-mandak
Super Member
This is not a yes or no thing; it's much more complicated than that. Simply do to how is up and over sampling done by the particular model, most major brands have there own way of doing this and therefore it is not possible to compare them as one.
Over sampling is as mentioned before in the thread just reading the same bit a number of times and then the logic decides if it is a "0" or "1", usually it will just be the number that gets most "hits" - no guarantee that it is the right guess.
Up sampling is much more complex and more likely to be wrong, since as cracker says it has to put information into the signal that is not there. Often quality is wrongly mistaken as a more comfortable sound. Most people can sense the sharp edges on the wav curve coming from an old CD-P (at least when comparing to a good quality vinyl cut of the same tune) but when they hear the CD on a new up sampling CD-P it is much more soft and nice for the ear. This is not quality but just making the wav curve smoother, there is no guarantee that the added information in the wav curve actually truly represents what was played in the studio - it is purely a result of the logic in the DAC. This is a fact and can not be discussed, the discussion can only be how close to the studio is the signal - is it close enough?
In the other end there is the no nonsense makers who says; do the job right from the beginning, don't use logic to save you're a.. Make a high quality transport that actually reads what is on the disk, and then there is no need for over sampling! Make a high quality DAC and analogue stage - then there is no need for up sampling.
In the Pro world the rule of thumb is simple; never convert or up sample if in any way possible. When sample rate is changed it is always downwards and in even numbers, if the recording is aimed for CD the recording is usually done in 88.2 kHz (2x44.1). Try do the math going from 96 and down to 44.1, bummer - remember there is no "after the comma". Now try to imagine going the other way from 44.1 and up to 96, you see how the problems just start piling up
This is not to speak for or against CD-P converting, but only to give an idea about what it's all about. In the end it will come down to the good old "In what do you believe, what is you're religion".
Over sampling is as mentioned before in the thread just reading the same bit a number of times and then the logic decides if it is a "0" or "1", usually it will just be the number that gets most "hits" - no guarantee that it is the right guess.
Up sampling is much more complex and more likely to be wrong, since as cracker says it has to put information into the signal that is not there. Often quality is wrongly mistaken as a more comfortable sound. Most people can sense the sharp edges on the wav curve coming from an old CD-P (at least when comparing to a good quality vinyl cut of the same tune) but when they hear the CD on a new up sampling CD-P it is much more soft and nice for the ear. This is not quality but just making the wav curve smoother, there is no guarantee that the added information in the wav curve actually truly represents what was played in the studio - it is purely a result of the logic in the DAC. This is a fact and can not be discussed, the discussion can only be how close to the studio is the signal - is it close enough?
In the other end there is the no nonsense makers who says; do the job right from the beginning, don't use logic to save you're a.. Make a high quality transport that actually reads what is on the disk, and then there is no need for over sampling! Make a high quality DAC and analogue stage - then there is no need for up sampling.
In the Pro world the rule of thumb is simple; never convert or up sample if in any way possible. When sample rate is changed it is always downwards and in even numbers, if the recording is aimed for CD the recording is usually done in 88.2 kHz (2x44.1). Try do the math going from 96 and down to 44.1, bummer - remember there is no "after the comma". Now try to imagine going the other way from 44.1 and up to 96, you see how the problems just start piling up
This is not to speak for or against CD-P converting, but only to give an idea about what it's all about. In the end it will come down to the good old "In what do you believe, what is you're religion".
Last edited:
morphcat
Member
Hi - Personally, I have yet to play around with the newer outboard oversampling DACS. However, I do have a Gigalab NOS DAC - essentially the same as the much hyped Valab DAC. Truth be told, I didn't really sit down and listen to it that much during the first 12 months I had it. Usually, when I can make time to REALLY listen to my system, I'm spinning vinyl.
Here's the thing - a couple of months ago I had the Pixie's playing from my computer through the Gigalab NOS while doing some house chores. At one point, I just happen to plop down in my sweet spot to tie my shoes. I immediately had one of those "holy sh*t" moments. I had NEVER heard the Pixies sound like that before. Unbelievable depth, separation and dare I say "vinyl/analog" sound.
I've since read similar descriptions of the "NOS Sound" from many others. All I can say is I am a believer. I'm still curious about some of the newer oversampling DACS - and one day I will get one to try. But for now - I listen to FLAC digital music with my NOS DAC almost as much as vinyl. It sounds REALLY good.....
Here's the thing - a couple of months ago I had the Pixie's playing from my computer through the Gigalab NOS while doing some house chores. At one point, I just happen to plop down in my sweet spot to tie my shoes. I immediately had one of those "holy sh*t" moments. I had NEVER heard the Pixies sound like that before. Unbelievable depth, separation and dare I say "vinyl/analog" sound.
I've since read similar descriptions of the "NOS Sound" from many others. All I can say is I am a believer. I'm still curious about some of the newer oversampling DACS - and one day I will get one to try. But for now - I listen to FLAC digital music with my NOS DAC almost as much as vinyl. It sounds REALLY good.....
Any competently-designed oversampling DAC will use a digital filter to interpolate between samples rather than simply repeating sample values. The Philips-designed players that ran the TDA1541 at a 4x oversampling rate always included the digital filter chip. There were some Philips players that used a TDA1541 at 1x in conjunction with a brick-wall op amp filter, but they were not popular with audiophiles because the brick-wall filters sounded bad.
StevenZ
Texas or nothin'
Great information in this thread. Just bought a TeraDak V3.1D which is a NOS DAC using 8x Philips TDA1543 chips. Claims 24/96 though as I understand it's a 16/44 chip. I'm hoping I like it since previously I've only had upsampling DACs.
I tried looking at the product description but I do not see any mention of a digital filter for this DAC. Any ideas?
I tried looking at the product description but I do not see any mention of a digital filter for this DAC. Any ideas?
osd1
Member
you would probably have to go to the chinesse website and google translate to "engwish". i read that project's dac's give you the option to either purchase a filtered or filterless dac.
i cant call myself a "purist" because i dont have enough knowledge about this stuff to claim that, but i am one to "walk away." i remember the first time i ever built a circular soffit. had to be dead balls on. as soon as we finished i said "wait shouldnt i just..." and he cut me off and said "kid, its ****ing perfect. dont touch it. put the screw gun down. you cant add another brace there is a ****ing pipe in the way. the soffits all may be the same, but yo gotta look at whats going on around you when putting them up." (give or take word for word) 9 years ago that has stuck with me.
moral of the trip down memory lane about my early days in construction is this: i think of the dac as the soffit and the mastering of recording as the ceiling we are attaching the soffit to.
as long as the dac is solid you are fine. as long as it functions and serves it purpose and does it with integrity and finesse and sounds great your good to go.
as far as if filters work, if upsampling works, its all dependent on each individual recording. you have 2 options, enjoy what your listening to, even if it has some short comings or set up 2 music servers or 2 cd players. one to a nos dac the other to a os dac.
pete_mac
Super Member
The TeraDak is filterless. I've heard the bigger brother, the TeraDak Chameleon which has 16x chips and it sounded really nice, if a tad challenged a the upper and lower extremes of the frequency range.
My current audio-gd DAC is fairly unique in that I can run it as NOS, 2x, 4x or 8x oversampling. I run it on the factory 8x but will experiment on a rainy day with the other oversampling and NOS settings.
charles 1973
Super Member
I have a Denon 1920 SACD/DVD Universal Player that Oversamples and Upsamples, I had an older Denon 1500II CD Player that Only Oversampled and it sounded better Overall that my Upsampling unit. Both units sound good overall though.
The Upsampling (and oversampling) DAC in my Bryston Integrated Amp Kills either of those two players (as it should for the price) in terms of SQ. It really is a Night and Day Difference on some disc's with respect to Imaging/Soundstage, Inner Detail and Cohearancy, on other disc's less so. My Friends Pioneer Elite P-D9 ($1,500) Upsampling/Oversampling SACD Player also sounded Great, very much like the Bryston DAC.
It seems like most other concepts in Audio, it depends on how well excuted the design is. My gut feeling is Oversampling is best with the High End Gear that has a Good Enough Audio Section to hear the difference. That is the area some CD Players skimp on and where No Amount of Oversampling can help. That said I believe the old standard of nothing more than 4x Oversampling, with No Upsampling, Properly Designed and Built is capable of Very Good Sound too.
The Upsampling (and oversampling) DAC in my Bryston Integrated Amp Kills either of those two players (as it should for the price) in terms of SQ. It really is a Night and Day Difference on some disc's with respect to Imaging/Soundstage, Inner Detail and Cohearancy, on other disc's less so. My Friends Pioneer Elite P-D9 ($1,500) Upsampling/Oversampling SACD Player also sounded Great, very much like the Bryston DAC.
It seems like most other concepts in Audio, it depends on how well excuted the design is. My gut feeling is Oversampling is best with the High End Gear that has a Good Enough Audio Section to hear the difference. That is the area some CD Players skimp on and where No Amount of Oversampling can help. That said I believe the old standard of nothing more than 4x Oversampling, with No Upsampling, Properly Designed and Built is capable of Very Good Sound too.
Last edited:
tppc
Dogs know best
@Netdewt, I would suggest exactly what pete-mac has if you are interested in exploring NOS vs OS. This particular DAC looks like it is packed with good engineering, features and future-proofing.
http://audio-gd.com/Pro/dac/RE5.2/RE5.2EN.htm
thymanst
Minnesota morning
I have a meridian 518 sample rate converter and has different filter slopes that can be changed ( it can change word size also) - every CD player has a filter slope to it. This is another variable that changes the sound of the dac. The new meridian gear has an apodizing filter which I've heard sounds excellent( I don't have any experience with it)
pete_mac
Super Member
The current Ref 5.32 is indeed a special DAC for the dollars.
In the Pro world the rule of thumb is simple; never convert or up sample if in any way possible. When sample rate is changed it is always downwards and in even numbers, if the recording is aimed for CD the recording is usually done in 88.2 kHz (2x44.1). Try do the math going from 96 and down to 44.1, bummer - remember there is no "after the comma". Now try to imagine going the other way from 44.1 and up to 96, you see how the problems just start piling up
I'd dispute that. In fact in the pro world there often is sampling rate and quantisation width conversion applied. Simply to reduce cumulative rounding loss in processing/mixing before the final downmix.
Greetings from Munich!
Manfred / lini
