[Ypoknons]

Estimates of costs from a DIY perspective based on cost of parts also don't take into account opportunity cost - i.e. the opportunity to make money you forego by taking the time to build the amp. Obviously this will be different for everyone - for students and those with more flexible schedules it might be $0; for those of us in jobs for a workaholics the amount can be quite substantial.

 

[Wharfrat]

Maverickronin wrote:
 
"Also, studies suggest that Redbook is indeed transparent to human ears unless you listen at insane volumes and need the extra dynamic range from 24 bits.  Hi-rez releases are just usually better mastered than normal CDs because anyone who spends the extra money on it is going to be more critical of the quality".
 
Transparent in the sense that the missing digital bits are not discernable to human hearing if one were to upsample from 44/16 to higher bit-sampling rates?  (Dan Lavry wrote a white paper arguing that anything above 96/24 was meaningless for all practical purposes, in the same vein that listening to sound above 15-20khz is.)
 
Or did you mean technically "transparent" in the sense there just more information on a high-res CD for a $30.00 (metaphorically speaking) CD-DAC to process without requiring that the unit contain a different set of internal components, such as more advanced DAC chipsets, to do so?  In other words, a laser reader is a laser reader is a laser reader and its no problem getting the digital bits into PCM waveforms and algorithmically computed to approximate the analog waveform to the point that missing audible information produced in the analog domain is not discernable?
 
You also piqued my interest: how could listening at insane volumes make something more discernable if listening for it at normal levels is not?  If its not audible at normal listening levels, wouldn't there be something wrong with the noise floor or the ability of the headphones to produce the "ghost" sound at normal listening levels? 
 
All I am saying here is that there is a price-performance curve where the audible components of a recording that are reproduced via source and transducer components share similar component costs and build labor costs leading to largely similar performance results...and that the price-performance ratio approaches zero somewhere in the $2000-$3000 range for standard CD players and arguably for most mid-sized monitors.  For headphones, the zero ratio point is somewhere between $800-$1200.  It is all subjective, but this is my take on the matter.
 
Still makes them expensive, given the build/sell costs vs. MSRP ratio of 1:5 in order to stay in business at a reasonable profit.  Beyond that, some sense of reasonableness needs to be maintained, otherwise the greed factor casts a tall shadow over the hobby.
 
 
 

 

[zhenya]

 
 
 


Quote:

Maverickronin wrote:
 
"Also, studies suggest that Redbook is indeed transparent to human ears unless you listen at insane volumes and need the extra dynamic range from 24 bits.  Hi-rez releases are just usually better mastered than normal CDs because anyone who spends the extra money on it is going to be more critical of the quality".
 
Transparent in the sense that the missing digital bits are not discernable to human hearing if one were to upsample from 44/16 to higher bit-sampling rates?  (Dan Lavry wrote a white paper arguing that anything above 96/24 was meaningless for all practical purposes, in the same vein that listening to sound above 15-20khz is.)
 
Or did you mean technically "transparent" in the sense there just more information on a high-res CD for a $30.00 (metaphorically speaking) CD-DAC to process without requiring that the unit contain a different set of internal components, such as more advanced DAC chipsets, to do so?  In other words, a laser reader is a laser reader is a laser reader and its no problem getting the digital bits into PCM waveforms and algorithmically computed to approximate the analog waveform to the point that missing audible information produced in the analog domain is not discernable?
 
You also piqued my interest: how could listening at insane volumes make something more discernable if listening for it at normal levels is not?  If its not audible at normal listening levels, wouldn't there be something wrong with the noise floor or the ability of the headphones to produce the "ghost" sound at normal listening levels? 
 
All I am saying here is that there is a price-performance curve where the audible components of a recording that are reproduced via source and transducer components share similar component costs and build labor costs leading to largely similar performance results...and that the price-performance ratio approaches zero somewhere in the $2000-$3000 range for standard CD players and arguably for most mid-sized monitors.  For headphones, the zero ratio point is somewhere between $800-$1200.  It is all subjective, but this is my take on the matter.
 
Still makes them expensive, given the build/sell costs vs. MSRP ratio of 1:5 in order to stay in business at a reasonable profit.  Beyond that, some sense of reasonableness needs to be maintained, otherwise the greed factor casts a tall shadow over the hobby.
 
 
 

 
Despite what is commonly believed in audio circles, the Red Book CD standard was developed from a very thorough understanding of what was necessary to reproduce the audible spectrum.  16 bits is sufficient to cover 96dB of dynamic range alone, and about 120dB with dithering.  If you consider that most recordings use as little as 10-20dB dynamic range, and even symphony recordings rarely use more than about 60dB,  and if we assume that the noise floor in a very quiet home is about 50dB, it is clear that even 96dB above that noise floor is more than enough to cause severe pain!
 
The 44.1k sample rate was chosen because is more than double the 20k maximum frequency they wished to be able to reproduce, and fully satisfies the Nyquist-Shannon theorem (with headroom to spare) that states that all that is required to perfectly re-create a signal is to have a sample rate double that of the frequency that one wishes to reproduce. 
 
So what do we get from a 24/96 recording?  The extra bits only give us more dynamic range (which can make the recording louder - but the 16 bits of CD can already theoretically get louder than we can physically tolerate) and more samples (but more samples won't produce a more accurate re-creation of the source, since 44.1k already is already an adequate sample rate.  You can't improve upon an already perfect re-creation).
 
See this thread http://www.head-fi.org/t/415361/24bit-vs-16bit-the-myth-exploded for further information.
 
 
 

 

[BlackbeardBen]

Quote:

Maverickronin wrote:
 
"Also, studies suggest that Redbook is indeed transparent to human ears unless you listen at insane volumes and need the extra dynamic range from 24 bits.  Hi-rez releases are just usually better mastered than normal CDs because anyone who spends the extra money on it is going to be more critical of the quality".
 
Transparent in the sense that the missing digital bits are not discernable to human hearing if one were to upsample from 44/16 to higher bit-sampling rates?  (Dan Lavry wrote a white paper arguing that anything above 96/24 was meaningless for all practical purposes, in the same vein that listening to sound above 15-20khz is.)
 
No, that downsampling higher resolution recordings to 16 bit, 44.1 kHz is inaudible with proper noise-shaping dither at normal listening volumes and otherwise normal listening conditions.  In some rare cases I would expect that ABX testing between the two may turn up differences that cannot be attributed to problems with the dither or equipment, but I would say that any possible differences are so low as to not make any practical difference at all.
 
Or did you mean technically "transparent" in the sense there just more information on a high-res CD for a $30.00 (metaphorically speaking) CD-DAC to process without requiring that the unit contain a different set of internal components, such as more advanced DAC chipsets, to do so?  In other words, a laser reader is a laser reader is a laser reader and its no problem getting the digital bits into PCM waveforms and algorithmically computed to approximate the analog waveform to the point that missing audible information produced in the analog domain is not discernable?
 
Regarding your last point; exactly.  It is not hard at all for very inexpensive modern DACs to reproduce the waveform represented digitally in PCM format near-perfectly in the analog domain.
 
Now, with high resolution audio the equipment actually becomes a limit - but that limit is so far down below the level of audibility it's a non-issue.  We're talking about the best DACs having an effective bit depth of 20 bits in analog output resolution, and the performance of 192 kHz sampling rates actually measuring worse than 96 kHz sample rates on DACs (in fact, the DAC-1 resamples everything to 110 kHz internally).
 
You also piqued my interest: how could listening at insane volumes make something more discernable if listening for it at normal levels is not?  If its not audible at normal listening levels, wouldn't there be something wrong with the noise floor or the ability of the headphones to produce the "ghost" sound at normal listening levels? 
 
It's not necessarily insane volumes so much (the ear's dynamic sensitivity ensures we wouldn't hear low-level details any better), but insane volumes for material that is relatively quiet on the recording.  So, imagine the sound of musicians in an orchestra turning pages of music between songs, with the recording done at a constant gain.  Under normal circumstances that sound is very quiet, but if you turn up the volume to what would be an insane level during a brass chorus, it would be pretty loud.  You're effectively using far fewer of the bits available - so the higher dither noise floor of 16/44.1 audio would become audible, unlike in normal listening conditions.
 
All I am saying here is that there is a price-performance curve where the audible components of a recording that are reproduced via source and transducer components share similar component costs and build labor costs leading to largely similar performance results...and that the price-performance ratio approaches zero somewhere in the $2000-$3000 range for standard CD players and arguably for most mid-sized monitors.  For headphones, the zero ratio point is somewhere between $800-$1200.  It is all subjective, but this is my take on the matter.
 
Ha!  $2000-3000?  Quite frankly, that's crazy.  Ignoring the transport side (i.e. looking at DACs only), based on objective performance metrics and testing, I think you hit that price-performance curve at $30 (Behringer UCA-202), with perhaps the most revealing systems (Stax, HD 800, etc.) potentially making distinctions possible between that and the $130 range (E-Mu 0204 or equivalent), which is about as objectively perfect in performance as possible short of the Benchmark DAC-1 (which is still far cheaper than your supposed price-performance zero point, even one you factor in a transport).
 
Informal case in point:
http://www.matrixhifi.com/contenedor_ppec_eng.htm
 
And a much more formal ABX test:
http://www.bostonaudiosociety.org/explanation.htm
 
Of course, deliberately induced colorations in frequency response, etc. are easily audible if present in more expensive DACs.  But any anomaly such as that is a flaw, no matter how good you think it sounds with the rest of the setup.  A DAC should be transparent and nothing else - Relatively speaking again, not a difficult or expensive feat.
 
Still makes them expensive, given the build/sell costs vs. MSRP ratio of 1:5 in order to stay in business at a reasonable profit.  Beyond that, some sense of reasonableness needs to be maintained, otherwise the greed factor casts a tall shadow over the hobby.
 

Oh, that greed factor casts a long shadow indeed.  Especially in the "tweak" category, which IMO encompasses high-end DAC voodoo.  I'm not talking just about the above-$2000 range, either.

 

 

[sphinxvc]

Quote:

 

[...]

Have you heard any top-end source?  There is definitely more to it than just FQ response from what I can tell.  Granted, all good DACs are more or less flat.  But take for example the Zodiac Gold.  I've heard one and know for a fact mine is inferior.  It may have just as flat a FR response, but that doesn't mean much, and I'm not really sure what the metrics are behind it, but it just simply sounds better (the gold)...and happens to cost $4K.
 

 

[tomb]

Quote:

Quote:

Have you heard any top-end source?  There is definitely more to it than just FQ response from what I can tell.  Granted, all good DACs are more or less flat.  But take for example the Zodiac Gold.  I've heard one and know for a fact mine is inferior.  It may have just as flat a FR response, but that doesn't mean much, and I'm not really sure what the metrics are behind it, but it just simply sounds better (the gold)...and happens to cost $4K.
 

This is absolutely correct.  Like others have said, it's relatively easy for any DAC to have a flat frequency response from 20 - 20KHz.  They draw the wrong conclusion from that, though.  Even minute differences in SN ratio, dynamic range and all the types of distortion get amplified all the way up the signal chain so that the differences are more than minor and can be easily heard.  It must also be remembered that once you have the DAC's output right at its Analog-out pins, all bets are off from that point forward.  All the issues of transmitting and amplifying an analog signal are in play. 
 

 

[mikeaj]

Quote:

This is absolutely correct.  Like others have said, it's relatively easy for any DAC to have a flat frequency response from 20 - 20KHz.  They draw the wrong conclusion from that, though.  Even minute differences in SN ratio, dynamic range and all the types of distortion get amplified all the way up the signal chain so that the differences are more than minor and can be easily heard.  It must also be remembered that once you have the DAC's output right at its Analog-out pins, all bets are off from that point forward.  All the issues of transmitting and amplifying an analog signal are in play. 

 
No, the argument is that you can get a flat frequency response AND extremely low distortion ("low" relative to what?  relative to what humans can perceive, and relative to headphones and speakers) AND low noise, and so on, at a fairly low cost.  Nobody except a fool is going to say that just the FR being flat is enough.  That's obviously wrong.
 
In high-end listening scenarios with great equipment, I wouldn't go so far as to say a $30 DAC can do everything sufficiently well, but it's probably a lot better than most expect.
 

 
 
 

 
Despite what is commonly believed in audio circles, the Red Book CD standard was developed from a very thorough understanding of what was necessary to reproduce the audible spectrum.  16 bits is sufficient to cover 96dB of dynamic range alone, and about 120dB with dithering.  If you consider that most recordings use as little as 10-20dB dynamic range, and even symphony recordings rarely use more than about 60dB,  and if we assume that the noise floor in a very quiet home is about 50dB, it is clear that even 96dB above that noise floor is more than enough to cause severe pain!

 
A very quiet home is probably in the 20s of dB, not 50 dB?  In the least, in the 30s (but then it would be a stretch to call it very quiet).
 
Also keep in mind that humans can discern sounds below a noise floor, quite easily.  But in practice, they can no longer do so when there's music playing that will obscure the really quiet details, so as you say, 96 dB is a LOT of dynamic range to work with and is plenty in realistic scenarios.  So for music playback where the level is not consistently really really low relative to 0 dBFS, 16-bit playback is just fine and dandy.

 

[BlackbeardBen]

Quote:

Quote:

Have you heard any top-end source?  There is definitely more to it than just FQ response from what I can tell.  Granted, all good DACs are more or less flat.  But take for example the Zodiac Gold.  I've heard one and know for a fact mine is inferior.  It may have just as flat a FR response, but that doesn't mean much, and I'm not really sure what the metrics are behind it, but it just simply sounds better (the gold)...and happens to cost $4K.
 

Yes, I have, although not extensively.  A Bladelius Embla ($10000 music server), and several high-end CD players in the $1000+ range.  Can't say I've ever heard anything out of the ordinary or "better" in any of them on various high-end systems compared to $300 CD players or even $150 DVD/BRD players - and that's on various high-end systems (my own included - Infinity Renaissance 90 speakers, a pair of Eico HF-12 monoblocks, and an Adcom GFA-555 for the woofers).
 
"Just sounds better"?  Let's see that it can be distinguished in a level-matched ABX test first against an E-MU 0204/0404 or equivalent (after ensuring it measures at least as well objectively - the specs mean nothing by themselves), and then if it passes let's do a blind test for preference.  Since you're so confident, that should be easy, right?  I'll eat my shirt if it does.
 
For reference, E-MU 0404 loopback test results:
http://www.amb.org/rmaa/
 
And that's combining the distortion for both output and input.  You think you can hear 0.0012% THD + Noise and 0.0014% IMD + Noise at normal listening levels, or that more than 113 dB dynamic range is audible at normal levels?  This is quite ridiculous, especially once you consider that even the best amplifiers don't perform this well - not to mention under difficult loads.
 
Of course, RMAA measurements don't tell the whole story - well for one, like I mentioned, the measurements I posted are through a loopback in the device itself.  Ideally you would use a superior input device so that its effects are minimized - but the only thing that that will do is lower the noise and distortion.  Also, more specific testing showing the actual characteristics of the harmonic and intermodulation distortion would be useful - but again, this is at levels such that none of it will be audible at all in the first place.

 
Quote:

This is absolutely correct.  Like others have said, it's relatively easy for any DAC to have a flat frequency response from 20 - 20KHz.  They draw the wrong conclusion from that, though.  Even minute differences in SN ratio, dynamic range and all the types of distortion get amplified all the way up the signal chain so that the differences are more than minor and can be easily heard.  It must also be remembered that once you have the DAC's output right at its Analog-out pins, all bets are off from that point forward.  All the issues of transmitting and amplifying an analog signal are in play. 
 

When those distortions and noise, inaudible as they are in any competent DAC, are amplified, so is the music signal the DAC delivers - masking them just the same.  The noise and distortion is additive, of course, with that added by amplifiers, etc.  But it's so low that it makes no discernible difference with competent DACs.

 

[buson160man]

  In reply to an earlier reply from obzilla.  I enjoyed your reply and am in total agreement with your line of thought.I have read about a lot of amps with envy knowing i could never justify spending that amount for them but dreams do not cost anything.But one day I just may pull the trigger and reach for the stars and for that moment at least I will have that ferrari and be in heaven at least for a little while or until someone comes up with another piece of headphone equipment that I will look at enviously.

 

[Bobsama]

If you're considering watts per channel dollars, I think you have a point. Ultimately, I think the real contribution is miniaturizing high-quality amps. It's almost cheaper to make a full-sized amp because the tolerances are a bit wider. A lot of boutique manufacturers need to design and build the circuit while paying for high-quality components. I don't know how much more "audiophile-quality" surface components cost but, as they're more expensive than off-the-shelf components, you can be sure that they'll further marked up because of the limited quantities being produced.

 

[leeperry]

I already opened a thread on this matter a while ago: http://www.head-fi.org/t/472214/are-most-headamps-overpriced-cmoys
 
Well, it makes perfect sense tbh: someone's gotta pay for R&D, mass production, RMA, labor taxes and employees wages, profit, allowing the resellers networks to make a living too, etc etc. And the less potential customers, the higher the price.
 
A famous headamp manufacturer told me that he had a 1:4 factor on all his gear and that the chassis itself already cost half of the production cost. The ppl who built a beta22 will tell you how much a good chassis costs, let alone proper stepped attenuators.

 

[tomb]

Quote:

Yes, I have, although not extensively.  A Bladelius Embla ($10000 music server), and several high-end CD players in the $1000+ range.  Can't say I've ever heard anything out of the ordinary or "better" in any of them on various high-end systems compared to $300 CD players or even $150 DVD/BRD players - and that's on various high-end systems (my own included - Infinity Renaissance 90 speakers, a pair of Eico HF-12 monoblocks, and an Adcom GFA-555 for the woofers).
 
"Just sounds better"?  Let's see that it can be distinguished in a level-matched ABX test first against an E-MU 0204/0404 or equivalent (after ensuring it measures at least as well objectively - the specs mean nothing by themselves), and then if it passes let's do a blind test for preference.  Since you're so confident, that should be easy, right?  I'll eat my shirt if it does.
 
For reference, E-MU 0404 loopback test results:
http://www.amb.org/rmaa/
 
And that's combining the distortion for both output and input.  You think you can hear 0.0012% THD + Noise and 0.0014% IMD + Noise at normal listening levels, or that more than 113 dB dynamic range is audible at normal levels?  This is quite ridiculous, especially once you consider that even the best amplifiers don't perform this well - not to mention under difficult loads.
 

RMAA is relative.  It's not the 113db dynamic range that's supposed to be audible.  It's whether you can hear more than a 2-3db difference with respect to another result from a different DAC.  The fact is, YOU CAN.
 
Quote:

 
Of course, RMAA measurements don't tell the whole story - well for one, like I mentioned, the measurements I posted are through a loopback in the device itself.  Ideally you would use a superior input device so that its effects are minimized - but the only thing that that will do is lower the noise and distortion.  Also, more specific testing showing the actual characteristics of the harmonic and intermodulation distortion would be useful - but again, this is at levels such that none of it will be audible at all in the first place.

 

I agree that the tested loopback should be superior.  However, once that has been established, other factors come into play.  As stated above, the absolute "zero" of RMAA measurement is entirely relative.  You can "trick" any result using RMAA if you're attempting to use it as an absolute scale, by tweaking the testing levels.  RMAA is of no value in that respect - only as a comparison with other test articles in as near identical conditions as possible.  Under that scenario, then the differences are noticeable at the 3db threshold and perhaps lower.
 
Quote:

When those distortions and noise, inaudible as they are in any competent DAC, are amplified, so is the music signal the DAC delivers - masking them just the same.  The noise and distortion is additive, of course, with that added by amplifiers, etc.  But it's so low that it makes no discernible difference with competent DACs.

Just curious, but how can you make a blanket statement such as that?  Seems to me that everything about audio and Hi-Fi is rolled up into that statement all by itself.  Knowing that RMAA is relative, any DAC with a difference in FR, Distortion, Noise, etc. of 3db or more compared to another DAC is going to be heard, period.

 
 

 

[Shike]

Quote:

Just curious, but how can you make a blanket statement such as that?  Seems to me that everything about audio and Hi-Fi is rolled up into that statement all by itself.  Knowing that RMAA is relative, any DAC with a difference in FR, Distortion, Noise, etc. of 3db or more compared to another DAC is going to be heard, period.

He never mentioned FR, just noise and distortion.  Also, I'd like you to provide a citation that shows audibility between -90dB and -87dB (for example) in relation to noise and distortion - since you claim it to be fact I'm sure you can find evidence of such a situation or similar correct?
 

 

[proton007]

I think the operational cost of a firm has been discussed thoroughly, but there are another two issues why things can be expensive:
 
a) A technological challenge, to prove a point. The Bugatti Veyron has little real world relevance. I mean it is impractical in most respects, and the amount of R&D put into it was more than the recovery, BUT it was done as a technical exercise, to prove that a car such as that can be built. Can the same be done by tuning an existing Corvette etc.? Maybe, but it was not *BUILT* for that. It never had that intention to begin with. So what is the ultimate value of the Veyron? It is the frontier of human intellect when it comes to personal transport, and you posses/enjoy *THAT* value if you buy a Veyron.
 
b) How much of a difference does it make *TO YOU*? A lot of people who own Porsche's seldom appreciate its real abilities, or they might not even have a use for those abilities. Yes, its fast in a straight line, but it is capable of a lot more, something that a Honda Civic cannot do.
 
I guess it comes down to this: You can be an audiophile with a lot of money, or you can be an audiophile. But you have to be one.
 

 

[tomb]

Quote:

He never mentioned FR, just noise and distortion.  Also, I'd like you to provide a citation that shows audibility between -90dB and -87dB (for example) in relation to noise and distortion - since you claim it to be fact I'm sure you can find evidence of such a situation or similar correct?
 

http://www.aes.org/e-lib/browse.cfm?elib=11981