Max,
The IIR filter is considered by many to be the best means of filtering for oversampling. It is more difficult to implement however... the process demands many more calculations. This is difficult, even at the relatively slow rate of audio.
Makes me wonder if implementing an IIR DAC wouldn't be a good a good project that could serve the DIY crowd. It would have to be worthwhile though. Being a dinosaur... I do little of this for the fun of it anymore.
Contentious perhaps... as this DAC thing seems to have become as nutty as speaker cables.
The IIR filter is considered by many to be the best means of filtering for oversampling. It is more difficult to implement however... the process demands many more calculations. This is difficult, even at the relatively slow rate of audio.
Makes me wonder if implementing an IIR DAC wouldn't be a good a good project that could serve the DIY crowd. It would have to be worthwhile though. Being a dinosaur... I do little of this for the fun of it anymore.
Contentious perhaps... as this DAC thing seems to have become as nutty as speaker cables.
rbflw, I thought I made my point clear (post #835). I'm not going to respond regarding this matter anymore.
poobah, I was just kidding about your age, but you are still my senior. So what if linear interpolation produces errors. Can it actually be heard and if so sound bad? If the DAC sounds good, I don't care.
poobah, I was just kidding about your age, but you are still my senior
. So what if linear interpolation produces errors. Can it actually be heard and if so sound bad? If the DAC sounds good, I don't care.
Hi poobah,
Hi MGH,
-Chris
You missed it, it is.
Hi MGH,
It's all relative. Your idea of what sounds good will adjust as you gain experience. My I suggest listening to live music for a while, with acoustic instruments. Then listen to your system. People claim to hear unmeasureable differences in speaker cables (I not going to get too far into poobah's favorite subject), and you wonder if you can hear problems in a data stream. Oh yes you can! See : http://www.diyaudio.com/forums/showthread.php?s=&threadid=84330 and a few other threads. I don't understand completely what poobah is saying, but I believe him. And I'm only 47!
-Chris
The question would be does the DAC sound better than what is already in your CDP.
The industry has perhaps stopped at sin x / x for filtering; it's quite clear that doesn't satisfy the audiophiles.
The truth be told... opting for 44.1 kHz for sampling was perhaps shortsighted or ill-conceived when Phillips & Sony put it all together. 80 kHz would have a better choice... but in 1982 that would have meant bigger discs or less minutes.
About things sounding "better"; I think we are too often fooled by something that sounds different. Chris makes excellent points about hearing live music. I played keys and bass to get through college... not for the money, but the girls. There is plenty that sounds "tizzy" and "harsh" in real music. If you want a serious case of listening fatigue; just stand next to a drummer for 3 hours. When people talk about soundstage and imaging... I am just on the floor laughing. We don't record music that way these days. We want REPRODUCTION that is accurate. This linear interpolation thing flies in the face of that.
I think EC is a pretty damn bright engineer as well. But the theory about sampling has been violated here... seriously.
If I hurt anyones feelings... sorry, but BS is BS, if someone doesn't know it's BS, then it's their fault. Read back all the way to the beginning, I started out gentle enough.
Again, learn the math.
The industry has perhaps stopped at sin x / x for filtering; it's quite clear that doesn't satisfy the audiophiles.
The truth be told... opting for 44.1 kHz for sampling was perhaps shortsighted or ill-conceived when Phillips & Sony put it all together. 80 kHz would have a better choice... but in 1982 that would have meant bigger discs or less minutes.
About things sounding "better"; I think we are too often fooled by something that sounds different. Chris makes excellent points about hearing live music. I played keys and bass to get through college... not for the money, but the girls. There is plenty that sounds "tizzy" and "harsh" in real music. If you want a serious case of listening fatigue; just stand next to a drummer for 3 hours. When people talk about soundstage and imaging... I am just on the floor laughing. We don't record music that way these days. We want REPRODUCTION that is accurate. This linear interpolation thing flies in the face of that.
I think EC is a pretty damn bright engineer as well. But the theory about sampling has been violated here... seriously.
If I hurt anyones feelings... sorry, but BS is BS, if someone doesn't know it's BS, then it's their fault. Read back all the way to the beginning, I started out gentle enough.
Again, learn the math.
poobah said:
They actually choose that earlier. In 1980 (IIRC -- i have a scan of the paper at home) Sony sent around a white paper. When it arrived i was taking the advanced math course that covered the process... i said they'll need at least 4 x to get it good enuff to challenge vinyl... examples of 96 & 192kHz sampled DVD-A go a long way to confirming that was a good call.
dave
Hello Chris and Poobah,
Regarding your posts:
"It's all relative. Your idea of what sounds good will adjust as you gain experience. My I suggest listening to live music for a while, with acoustic instruments. Then listen to your system." -Chris
"About things sounding "better"; I think we are too often fooled by something that sounds different. Chris makes excellent points about hearing live music." -Poobah
Actually I've been playing the classical guitar since I was a teenager and dabbled in playing the sax and piano. I try to catch performances by the Boston Symphony Orchestra as much as I can. So I have an idea of what good live acoustic music sounds like. When I ask if something sounds good or better, I use live acoustic music and my experience as an amateur musician for reference. Is that good enough for you?
Regarding your posts:
"It's all relative. Your idea of what sounds good will adjust as you gain experience. My I suggest listening to live music for a while, with acoustic instruments. Then listen to your system." -Chris
"About things sounding "better"; I think we are too often fooled by something that sounds different. Chris makes excellent points about hearing live music." -Poobah
Actually I've been playing the classical guitar since I was a teenager and dabbled in playing the sax and piano. I try to catch performances by the Boston Symphony Orchestra as much as I can. So I have an idea of what good live acoustic music sounds like. When I ask if something sounds good or better, I use live acoustic music and my experience as an amateur musician for reference. Is that good enough for you?
poobah said:
True.
Look at 15 kHz. K2 is 30 kHz.
Look at 12 kHz. K2 is 24 kHz.
Look at 10 kHz. K2 is 20kHz K3 is 30 kHz.
Look at 8 kHz. K2 is 16kHz K3 is 24 kHz.
All those harmonics are overtones that fill the empty space on the CD format, they enhance the limited CD response to 30 kHz ore more if it is a filterless or low order filtered DAC.
That could be a very positive aspect of linear interpolation.
Look at 6 kHz. K2 is 12 kHz which is almost an overtone and also K2 is a nice harmonic. K3 is already 18 kHz and overtone. Also here we already have some more sample points, distortion goes down.
Look at 4 kHz. K2 is 8 kHz, nice harmonic, no problem, K3 is 12 kHz, almost an overtone. More samplepoints, lower distortion.
Look at 2 kHz. K2 is 4 kHz and nice. K3 is 6kHz, acceptable. K4 is 8 kHz, even order, still acceptable. K5 is 10 kHz. This is already a higher order odd harmonic that could be disturbing, 10 kHz is not yet an overtone. But even more samplepoints and less distortion.
Look at 1 kHz. K5 is 5 kHz and the first ugly harmonic but again more samplepoints and lower distortion.
IMHO the area above 5 kHz is not a problem, even more it could enhance the bandwith-limited CD format. It could reconstruct the information above 20 kHz that was lost in the sampling process.
The question is, if distortion gets noticeable between 500 Hz and 5 kHz ?
Are there enough samplepoints ?
Maybe somebody can run a fft simulation of 8x linear interpolated signals at 1 kHz, 2,5 kHz and 5 kHz ?
Anyway my impression of standard non os with low order filter is that it sounds more open and transparent compared to os with same DAC chip.
Whatever the reason is...
Thanks, Poobah
More stuff to study...
OT: I found interesting the DAC master-clocked->CDP slaved concept. It is said to sound good.
Yes, little town, few good artists.
I play human larinx though
All I want is a natural sounding DAC with no HF noise
(at 39, I still hear 20Khz...or at least I think so)
M
Hi Chris,
"I still can not fathom how a source with digital errors could possibly sound better to you. That flies in the face of all my CD or digital source experience. There must be other factors at play." -Chris
I'm not sure if you're referring to linear interpolatin - I've never heard this so I can comment. But NOS that I have heard sound more natural to me. I used to own the Ack! DAC, a battery powered NOS TDA1545 DAC. This DAC sounded the most natural for vocals, sax, and piano compared to all other OS designs I've heard or owned. But it did not sound as open or "sparkly" as some OS designs which may be due to the natural HF roll off of NOS design. I'm hoping EC's Octal DAC will improve on the short commings of most NOS designs. I currently own a OS CDP which I thought sounded better than Ack! DAC when I bought it, but as time went by I started to miss the more natural sound of the Ack! - wish I hadn't sold it.
I've studied human perception (vision and hearing) during my PhD thesis years, and it is a complex subject. Just because we can't measure something (whether it's sound of CDP, cables - Poobah control yourself, etc.), doesn't mean it doesn't exist. We just haven't learned to measure the right parameter. An example is how ultrasonics (eg, 30 kHz), sound we humans aren't suppose to hear, can modify our perception of lower audible HF sounds. I don't remember the exact publications, but one of the seminal work came out of Cal Tech awhile back.
I think the problem with many engineers is that they have rigorously trained in measuring particular set of parameters in audio that may not be that important to human perception of sound. For example, is a DAC's ability to reproduce a perfect 18 kHz sine wave important, when in reality we hear a complex waveform in music? I personally find the sound of a perfect 18 kHz sound wave very boring - yes, I still can hear that high. I'm simplifying things, of course, to get my point across. I don't have the right answers, but I don't think high-end audio engineering gurus have the right answers either.
I think what EC is doing is as rigorous as one man can be designing a DAC from ground up. His choice of digital sampling may fly in the face of conventional wisdom, but he lets his hearing, not theory, guide his design which is the common and most important endpoint IMHO. Other designers (Doede etc) has followed this sensible approach as well.
"I still can not fathom how a source with digital errors could possibly sound better to you. That flies in the face of all my CD or digital source experience. There must be other factors at play." -Chris
I'm not sure if you're referring to linear interpolatin - I've never heard this so I can comment. But NOS that I have heard sound more natural to me. I used to own the Ack! DAC, a battery powered NOS TDA1545 DAC. This DAC sounded the most natural for vocals, sax, and piano compared to all other OS designs I've heard or owned. But it did not sound as open or "sparkly" as some OS designs which may be due to the natural HF roll off of NOS design. I'm hoping EC's Octal DAC will improve on the short commings of most NOS designs. I currently own a OS CDP which I thought sounded better than Ack! DAC when I bought it, but as time went by I started to miss the more natural sound of the Ack! - wish I hadn't sold it.
I've studied human perception (vision and hearing) during my PhD thesis years, and it is a complex subject. Just because we can't measure something (whether it's sound of CDP, cables - Poobah control yourself, etc.), doesn't mean it doesn't exist. We just haven't learned to measure the right parameter. An example is how ultrasonics (eg, 30 kHz), sound we humans aren't suppose to hear, can modify our perception of lower audible HF sounds. I don't remember the exact publications, but one of the seminal work came out of Cal Tech awhile back.
I think the problem with many engineers is that they have rigorously trained in measuring particular set of parameters in audio that may not be that important to human perception of sound. For example, is a DAC's ability to reproduce a perfect 18 kHz sine wave important, when in reality we hear a complex waveform in music? I personally find the sound of a perfect 18 kHz sound wave very boring - yes, I still can hear that high. I'm simplifying things, of course, to get my point across. I don't have the right answers, but I don't think high-end audio engineering gurus have the right answers either.
I think what EC is doing is as rigorous as one man can be designing a DAC from ground up. His choice of digital sampling may fly in the face of conventional wisdom, but he lets his hearing, not theory, guide his design which is the common and most important endpoint IMHO. Other designers (Doede etc) has followed this sensible approach as well.
Hi MGH,
The digital errors can be transport generated, or any part of the digital chain which will include demodulation, filtering (I think poobah is commenting on this) or in the conversion process to analog.
I believe we agree on principle then. The reason a non-os D/A sounds like it has a natural roll off is probably the vicious filter than must come after it. Your image is going to start appearing just after 22 KHz. It has to be chopped off. If you think of it, the bulk of sound differences are probably what the designer does after the conversion.
Now, if you oversample, you can move those images way out. The reason you want to do that is so you can use a filter that has a much nicer phase response. More Bessel oriented. Now you do not have the phase changing rapidly, or any ripples in the amplitude response. What you do in the digital domain is a little beyond my understanding, but that will have effects back in our passband unless they are very careful about what they do.
-Chris
The digital errors can be transport generated, or any part of the digital chain which will include demodulation, filtering (I think poobah is commenting on this) or in the conversion process to analog.
I believe we agree on principle then. The reason a non-os D/A sounds like it has a natural roll off is probably the vicious filter than must come after it. Your image is going to start appearing just after 22 KHz. It has to be chopped off. If you think of it, the bulk of sound differences are probably what the designer does after the conversion.
Now, if you oversample, you can move those images way out. The reason you want to do that is so you can use a filter that has a much nicer phase response. More Bessel oriented. Now you do not have the phase changing rapidly, or any ripples in the amplitude response. What you do in the digital domain is a little beyond my understanding, but that will have effects back in our passband unless they are very careful about what they do.
-Chris
hi john, ca U tell me that do U like I reserve some of same date code chips for U to prepare for your project??
Because one diy website ask me to order one lot of chips for them for their 4 pcs 1541a dac.
I will first priority for U if U still need the chips.
Pls let me know asap.
thx
thomas
Because one diy website ask me to order one lot of chips for them for their 4 pcs 1541a dac.
I will first priority for U if U still need the chips.
Pls let me know asap.
thx
thomas
Hi all,
After some negative comments on my octal D-I DAC project and this thread, even from a diyAudio Moderator (as this DAC thing seems to have become as nutty as speaker cables > You missed it, it is) I am baffled. I removed my signature and avatar befor I get comments on them too. This is a diyAudio site right? it's about people building audio projects right? Well that's exactly what I have done, I documented the DAC project I am currently working on and described it in my own way, then it is called nutty. All is carefully backed-up with measurements and listening tests.
DiyAudio members have come over to listen to the DAC to verify it's sound quality, what more could I have done?
The perceived sound doesn't have the high frequency interference that's often generated by NOS DAC's. The high frequency response is a lot better than many people think. Because it's still a NOS DAC, it's relatively immune to jitter (178ps) when compared to 8 times oversampled DAC's (1.35ps), it has a differential I2S low jitter interlink as standard. A automatically switching dual format circuit accepts both Philips and Sony format directly from a transport's I2S outputs. It also has a integrated USB interface with a unique reclocker that produces amazing sound quality for a computer based sound source. De DAC's analog output circuitry with very high initial output voltages makes sure noise levels are extremely low, despite the components used. It has the unique capability to switch between 3 output modes: tube, op-amp and mixed mode.
My main reason for starting this thread was to document a diyAudio project from scratch. Hoping it would inspire people to build their own audio equipment.
I am 100% convinced the DAC documented on this thread is something very special, and is certainly not just another NOS DAC. Every aspect of this DAC is tested and optimized in practice. Just have a look at all the effort I put in this project to optimize the sound quality.
During listening tests high quality chesky audio recordings were used that produce very natural sound quality. The sonic resonators used are like magnifying glasses put on the audio signal, they reveal everyting, every little flaw in the audio chain is immediately audible. This is the environment the octal D-I DAC has been tested in, day after day for about half a year. Now think how this would reflect on the sound quality it produces.
I personally thank all diyAudio members who have supported me. I will continue to do my very best to make the octal D-I DAC even better.
After some negative comments on my octal D-I DAC project and this thread, even from a diyAudio Moderator (as this DAC thing seems to have become as nutty as speaker cables > You missed it, it is) I am baffled. I removed my signature and avatar befor I get comments on them too. This is a diyAudio site right? it's about people building audio projects right? Well that's exactly what I have done, I documented the DAC project I am currently working on and described it in my own way, then it is called nutty. All is carefully backed-up with measurements and listening tests.
DiyAudio members have come over to listen to the DAC to verify it's sound quality, what more could I have done?
The perceived sound doesn't have the high frequency interference that's often generated by NOS DAC's. The high frequency response is a lot better than many people think. Because it's still a NOS DAC, it's relatively immune to jitter (178ps) when compared to 8 times oversampled DAC's (1.35ps), it has a differential I2S low jitter interlink as standard. A automatically switching dual format circuit accepts both Philips and Sony format directly from a transport's I2S outputs. It also has a integrated USB interface with a unique reclocker that produces amazing sound quality for a computer based sound source. De DAC's analog output circuitry with very high initial output voltages makes sure noise levels are extremely low, despite the components used. It has the unique capability to switch between 3 output modes: tube, op-amp and mixed mode.
My main reason for starting this thread was to document a diyAudio project from scratch. Hoping it would inspire people to build their own audio equipment.
I am 100% convinced the DAC documented on this thread is something very special, and is certainly not just another NOS DAC. Every aspect of this DAC is tested and optimized in practice. Just have a look at all the effort I put in this project to optimize the sound quality.
During listening tests high quality chesky audio recordings were used that produce very natural sound quality. The sonic resonators used are like magnifying glasses put on the audio signal, they reveal everyting, every little flaw in the audio chain is immediately audible. This is the environment the octal D-I DAC has been tested in, day after day for about half a year. Now think how this would reflect on the sound quality it produces.
I personally thank all diyAudio members who have supported me. I will continue to do my very best to make the octal D-I DAC even better.
John, you did so much in this 1/2 year, i modified only a player in this time. Its not only the work on it, but you did a great job, also in time consuming listening tests and reports, plus all answers here on the forum. All beside daytime job. What could be said too is: you always kept polite, no matter critics.
Maybe its time now to take some rest, sit back, listen to the sound of YOUR musical creation and laugh about all others with their comments. Serving all members their needs is impossible. The audio community has too much "This is how it's supposed to be" people, thats what i know now. But the popularity of this thread shows you have about 99.99 % supporting members.
Maybe its time now to take some rest, sit back, listen to the sound of YOUR musical creation and laugh about all others with their comments. Serving all members their needs is impossible. The audio community has too much "This is how it's supposed to be" people, thats what i know now. But the popularity of this thread shows you have about 99.99 % supporting members.