Telstar, they don't sound fake to me. But a NOS sound is terrible for MY ears.
The conviction that NOS is better it is a matter of "religion" I think, that's why I don't try to convince nobody of nothing.
To MY ears, NOS sounds coloured, terrible, artificial soundstage, OS 4x followed by a 16 bit DAC sounds somehow acceptable (but not easy) and OS 16x followed by a good quality 24bit DAC sounds better.
The ultimate sound for me it is SACD and DVD-A - all my new records are in one of those two formats.
Also computer sound for me it is also horrible. The digital noise level inside a PC case, the noise of the switched-mode power rails shared with everything, jitter because of the spread-spectrum clock that 99% it is left on by default in BIOS.
USB it is even worse when it is about jitter.
Fireware it is somehow a better solution, but still, the fan noise itself...
Numbers mean something, don't fool yourself. The fact that do not mean everything doesn't make true the myth that they do not mean absolutelly nothing. It is a faulty logic.
PS: I have a modded player with TDA 1541 that I can switch between NOS and OS in real time. And I have listened one of the NOS DAC's that are found on eBay - not too much of a difference compared to my mod - IMHO.
The conviction that NOS is better it is a matter of "religion" I think, that's why I don't try to convince nobody of nothing.
To MY ears, NOS sounds coloured, terrible, artificial soundstage, OS 4x followed by a 16 bit DAC sounds somehow acceptable (but not easy) and OS 16x followed by a good quality 24bit DAC sounds better.
The ultimate sound for me it is SACD and DVD-A - all my new records are in one of those two formats.
Also computer sound for me it is also horrible. The digital noise level inside a PC case, the noise of the switched-mode power rails shared with everything, jitter because of the spread-spectrum clock that 99% it is left on by default in BIOS.
USB it is even worse when it is about jitter.
Fireware it is somehow a better solution, but still, the fan noise itself...
Numbers mean something, don't fool yourself. The fact that do not mean everything doesn't make true the myth that they do not mean absolutelly nothing. It is a faulty logic.
PS: I have a modded player with TDA 1541 that I can switch between NOS and OS in real time. And I have listened one of the NOS DAC's that are found on eBay - not too much of a difference compared to my mod - IMHO.
This discussion is becoming a little pointless . Everyone here discusses the validity of OS vs. NOS. Some favour the first, others the latter.
Why is that? Forget about OS vs NOS, think about how different hearing is from person to person. Some may be sensitive to phase shifts, others may detect the distortion of aliasing better, and so on.
Believe it or not, but I once had a teacher who claimed he could NOT tell if speakers were connected in-phase or out-of-phase! For me that is very hard to believe, but I'm willing to accept the possibility he actually couldn't.
Another expample: when I was younger I could tell instantly (i.e. without a-b comparison) whether I was listening to lossy compression or to uncompressed, while most others could not. For this fact I was actually invited by Philips to one of their service centres to listen to some more upmarkted DCC-recorders than the one I had returned to have repaired (little did I know at the time that it wasn't the fault of that recorder, they all did it).
There are probably many more differences like this... So whether you'll like NOS or OS is up to you to find out.
Why is that? Forget about OS vs NOS, think about how different hearing is from person to person. Some may be sensitive to phase shifts, others may detect the distortion of aliasing better, and so on.
Believe it or not, but I once had a teacher who claimed he could NOT tell if speakers were connected in-phase or out-of-phase! For me that is very hard to believe, but I'm willing to accept the possibility he actually couldn't.
Another expample: when I was younger I could tell instantly (i.e. without a-b comparison) whether I was listening to lossy compression or to uncompressed, while most others could not. For this fact I was actually invited by Philips to one of their service centres to listen to some more upmarkted DCC-recorders than the one I had returned to have repaired (little did I know at the time that it wasn't the fault of that recorder, they all did it).
There are probably many more differences like this... So whether you'll like NOS or OS is up to you to find out.
If I may, I would comment that the technically correct solution is OS, since no real-world analog filter can block the aliasing products of 44.1kHz conversion without affecting major part of the audio spectrum. But from the sonical point of view, it really depends on personal preferences of a listener.
Non os is technicall correct.
Digital Filters are never lossless.
For comparison, I once owned two technical video cameras, one had analog gain, the other was a kind of mkII version that had digital signal processing, everything else beeing equal.
The analog camera had some noise on max. gain.
The DSP camera had overall inferior picture quality with grain and loss of sharpness on increased gain.
By the way, TDAs are not very suitable to judge sound of nonos.
Clearly increased distortion.
Digital Filters are never lossless.
For comparison, I once owned two technical video cameras, one had analog gain, the other was a kind of mkII version that had digital signal processing, everything else beeing equal.
The analog camera had some noise on max. gain.
The DSP camera had overall inferior picture quality with grain and loss of sharpness on increased gain.
By the way, TDAs are not very suitable to judge sound of nonos.
Clearly increased distortion.
SoNic_real_one said:
See below.
SoNic_real_one said:
It's not a religion for me. Actually, I couldnt care less, as long as it sound better.
But I came to the conclusion that the oversampling algorithms are not transparent. They cant be with the current processing power. If you noticed my other posts here and on the asylum, you know which is my answer. Software oversampling, processed with 64bit precision and the least damaging algorithm (which is supposed to be 0 stuffing, but I'm not sure yet). But that requires a computer (and a powerful one), sorry.
Bernhard said:
This is the answer. the bitrate loss is too much to process 16bit material, as it has been said in this very thread.
So if the only NOS comparison is one of the many cheap tda1543 DACs sold on ebay, it's not a fair comparison.
Sonic, Try an audionote dac3, a naim cds3x, a lector digicode or a lessloss dac2004. Then let's see if your judgement stand still.
Well, true. Maybe the TDA1541 it is at fault somehow. But I doubt that that's all it's fault...
Sure, OS has also it's ways to be done wrong (like in the jittery SAA companion of TDA1541) or well (like in latest 24bit DAC's). I was talking about the best-case scenario, with the OS done correctly 🙂 I don't see why the interpolation process it needs more than extra 4 bits for as 16x OS thou. That's it if no resampling is done and linear interpolation (which I think it's all you need)...
If you resample or want any other algorithm or interpolation, then yes, a 32 bit path is needed.
Sure, OS has also it's ways to be done wrong (like in the jittery SAA companion of TDA1541) or well (like in latest 24bit DAC's). I was talking about the best-case scenario, with the OS done correctly 🙂 I don't see why the interpolation process it needs more than extra 4 bits for as 16x OS thou. That's it if no resampling is done and linear interpolation (which I think it's all you need)...
If you resample or want any other algorithm or interpolation, then yes, a 32 bit path is needed.
SoNic_real_one said:
Nope, it can be anything from power source to the i/v stage. 🙂
Multibit DACs are very sensitive.
SoNic_real_one said:
That works only for delta-sigma dacs, which are not my cup of tea.
SoNic_real_one said:
Linear interpolation fascinates me, it was the first thing I thought about
But I'm afraid of the HF rolloff. You sure have seen the 4x tda1541a linear interpolation project by ec-designs on this very forum. I was thinking of a similar approach with pcm1704, and I still havent got that over my head 🙂
But now I believe that software upsampling (with 64bit precision) to NOS 24bit* DAC can be an easier and maybe better choice.
[*here too, paralleling at least 4 pcm1704 per channel should preserve 22 or 23 real bits.]
I think too that is a cheaper and cleaner solution. Maybe the DSP that will do the linear interpolation will be better at that than what's inside of a DAC. The only thing that matters it is to push the image far enough to be easily and clean filtered.
I saw the project with 4 DAC's and IMHO that's a brute mode to achieve that interpolation but I don't think they emulate well the necessary extra 2 bits (for 4x - 4 DAC's).
I saw the project with 4 DAC's and IMHO that's a brute mode to achieve that interpolation but I don't think they emulate well the necessary extra 2 bits (for 4x - 4 DAC's).
rtate said:
Since into the question you mention computer audio I would say that the computer and Operative Sistem you will use will make the big first difference .
Personally with the same laptop and same audio hardware I cant stand the audio coming from a Windows Vista , I believe it is a matter of jitter , latence and to much processing maybe .
While with the Ubuntu studio and Jack Audio ( with or without a realtime kernel) I am -conversely- enough pleased with the sound to not believe it is coming from a computer.
My listening is expecially good when compared to usb audio (and also the internal audio board do wonders when connected through Jack Audio) . For Usb audio I use an EMU0404 and you know what DAC there is inside it . It plays beautifully correct through Ubuntu studio/ Jack audio . Once said that , I am convinced that also a multibit dac can go very good with it . One big thing is important for me : correct phase between channels and image rejection .That means also correct soundstage and natural dynamics .The soundstage -of course- can be damn wrong and bad on both OS or NOS when the source is not good ( I refer to many cd player around ).On the Os or NOS thing , I tend to agree with the member SoNic_real_one about the effect of the NOS on my ears even if Multibit DACS can show more soul to the music at times .Dont want to go into the better-NOS-dac-design-around matter becouse the major thing about reproduced sound -by my point of view - is the ability to render the soundstage credible and possibly- also a credible phase coherence between instruments .Just go to teathers once in a while .A soundstage that you dont need 2 brains to stay in front of , and of course to merge inside . And to make sound decently a NOS is not an easy task IMHO .
Ah , if you can ,chek out audio polarity correctness everytime .
I think vista has a problem with DPL latency. I have not really noticed degradation in audio or anything because of it but the Latency floor is much higher than XP and the potential for it to get worse and spike is there.
SoNic_real_one said:
I bypassed the SAA7220 and the 1541A sounded extremely coloured, the plain 1541 was even worse.
The 1543 is crap anyway.
[*here too, paralleling at least 4 pcm1704 per channel should preserve 22 or 23 real bits.]
😀 or it just wont work .
http://www.gte-audio.com/pdf/lianotec_d.pdf
look they brag about "chips direct from factory" and bout 21.
😀 or it just wont work .
http://www.gte-audio.com/pdf/lianotec_d.pdf
look they brag about "chips direct from factory" and bout 21.
Bernhard said:
That's what happend with my plain 1541 in NOS mode. But I don't think it is the DAC fault but of the not bricked images.
There it is paralleling only on the analog side, the input is linear OS. Their results of 8x are pretty 🙂tritosine said:
yes but(!!!) a bog standard pcm1704 "indicates 16.5 bit inherent linearity" . Those used in those measurements are therefore not standard but handpicked at factory, this isnt diy anymore.
I tend to agree with Bernhard BTW.
DBX Aes paper 1985 :
"In addition to the measurements listed above, informal
listening tests were conducted at a local studio, using live
source material and comparing the output of the A/D-D/A chain
with the live signal. In single-blind tests, no one could
reliably distinguish between the processed and unprocessed
signal."
AD 6bit front end noiseshaping 18bit, 48khz output
DA 16 (+2 time averaged) 18bit, 48khz input- "The lowest two bits are decoded to determine the correct time at which to increment the counter. The D/A output is then de-glitched and applied to a sin(X)/X correction filter and output-reconstruction filter of conventional design."
Author : Robert W. Adams
DBX Aes paper 1985 :
"In addition to the measurements listed above, informal
listening tests were conducted at a local studio, using live
source material and comparing the output of the A/D-D/A chain
with the live signal. In single-blind tests, no one could
reliably distinguish between the processed and unprocessed
signal."
AD 6bit front end noiseshaping 18bit, 48khz output
DA 16 (+2 time averaged) 18bit, 48khz input- "The lowest two bits are decoded to determine the correct time at which to increment the counter. The D/A output is then de-glitched and applied to a sin(X)/X correction filter and output-reconstruction filter of conventional design."
Author : Robert W. Adams
tritosine said:
I dont speak german and I'm not sure what you are reffering to in that document.
If you mean that I would need 8 DACs, it has to be seen if the Trinity uses 8 dacs in total or 8x channel. I was referring per channel of course, because the pcm1704 is mono.
I dont know what "it wont work", but that writeup shows the bads done by analog filtering pretty clearly.
tritosine said:
http://www.gte-audio.de/pdf/lianotec_e.pdf
Found english version. Will read it later.
Well, for the pricetag they can handpick 1/1000 DACs 😉
This is a more interesting read:
http://www.nalanda.nitc.ac.in/industry/AppNotes/BurrBrown/appnotes/DesignSem5.pdf
- Status
- Not open for further replies.