[bigshot]

Do you know the threshold of audibility for this spec? How much is enough?

 

[yadako]

writing this up just to appreciate the very 1st thread! thanks for the info

 

[Otodynia]

Do you know the threshold of audibility for this spec? How much is enough?

I'm not sure I understand your question, but I can try to explain my reasoning (which I think has been shaped by earlier discussions in this thread):
 
If the full 96dB dynamic range of the 16/44.1 format is sufficient to blow out my eardrums, why should I worry about whether a DAC itself has a noise floor of 105 dB or 120 dB?  (I've shifted my thinking a bit since my last response to you, apparently.)
 
Was that what you were asking?  If not, then I will ask you back: what is the audibility threshold of the spec?  Is my thinking wrong?
 
The question from my response to castleofargh still stands, though: is there really such a thing as a "bad" DAC these days?  I'm not talking about the more esoteric botique brands (since I'm not ever going to spend that much money), but it seems that the larger, established companies can hire people who really know what they're doing when it comes to the digital-to-analog conversion process.  Microscopic evaluation of the output waveforms might reveal this or that, but am I going to really hear the difference?  Even if I can, how am I, the consumer, supposed to know which microscopic difference is right and which is wrong?

 

[bigshot]

That's exactly what I was driving at.

And an even better question than your "bad DAC" question is, are most players audibly transparent even without an external DAC?

 

[Otodynia]

That's exactly what I was driving at.

And an even better question than your "bad DAC" question is, are most players audibly transparent even without an external DAC?

That's a question that my limited knowledge can't answer.  What's your opinion?  
 
Having thought about why I was going to look at DAC SNR specs, I can say that I was using at it more from the standpoint of "engineering quality" (i.e., whether the manufacturer did a good job or not) rather than "sound quality."  
 
Really, though, it just doesn't seem to matter that much.  At this point, I'm more inclined to just compare features (how many formats are decoded, does it have a USB input, etc.) to determine if a DAC/DAP/streamer will meet my needs rather than nitpick the technical specs.  Is that misguided?
 
Or, if you'll tolerate another question, if you were shopping for a DAC, what would you look for?  

 

[bigshot]

Just about all solid state electronics (DAPs, CD/DVD/Blu players, DACs / amps) are audibly transparent. This means they all sound the same. There are differences in the way they measure, but who cares if you're listening to music?

Do you need a high end DAC? Do you need 24 bit? Do you need to spend a lot of money? Nope.

If I was shopping for a DAC, I would look for a player that didn't require an external DAC. The same goes for amps. I's look for headphones that didn't require amping. I'd take the money I saved and spend it on better headphones.

 

[cjl]

Many (but certainly not all) portable digital players are audibly transparent, including many cell phones. There are a couple of common problems to look for though. The two I have personally noticed with some players are an audible noise floor and significant output impedance. An audible noise floor is easy to detect - play a silent audio file, or an audio file with a silent passage at the maximum volume you would ever listen to the player using your most sensitive set of headphones. If you can't hear any noise or hiss, it isn't a problem. As for the output impedance, that can be harder to find. With some headphones, this isn't an issue (dynamic IEMs don't tend to have a problem with this, for example), but with other headphones, it can be a significant problem. With balanced armature IEMs, a high output impedance can cause significant distortions in the frequency response, due to the significant variation in the impedance of the IEM with frequency, and with some low-impedance headphones that use electrical damping to control the driver at low frequencies (for example the older Denon AH-Dx000 series), a high output impedance can cause the bass to become boomy and distorted.
 
Unfortunately, unless you have equipment like an oscilloscope or at least a good AC voltmeter, you have to detect this problem either by ear (by comparing its output to a known good source) or by googling your player and hoping someone else has measured the output impedance. If you do have access to a good AC voltmeter or scope though, you can put a test signal through your device and measure the output voltage with no load. Then you can put a 16 ohm resistor (or similar) across the output of the device and measure the voltage again. If it has a very low output impedance, the level should be basically identical for both the unloaded and loaded case. If your device has a 16 ohm output impedance though (as an example), it will put out half as much voltage with a 16 ohm resistor across its output as it will when unloaded, and this could cause audible issues with some (but not all) headphones.

 

[bigshot]

Or just get headphones where impedance isn't an issue.

 

[cjl]

Or just get headphones where impedance isn't an issue.

Sure, except that I haven't found ones I like as much as my D5000s, and they have an issue with a high source impedance. Besides, any competently designed headphone output should have a low impedance (but sadly, many aren't competently designed). I end up just running an O2 amp/USB dac most of the time, since I primarily listen to headphones at work, and my work computer has the most pitiful headphone output you've ever heard (full volume is rather quiet with my D5ks, which aren't exactly power hogs, and it has a fair amount of hiss too). The O2 has an inaudible amount of noise and an output impedance of significantly below 1 ohm, so for my purposes, it's audibly perfect.

 

[bigshot]

Amping isn't as much of an inconvenience for a home or office system I suppose. Worse for a portable system.

 

[castleofargh]

 

  SNR if given with no information on how it's done is mostly a commercial trick. from what I've seen about it, you can use several methods that won't give the most flatering results, or you can just filrter out the frequencies of shame to have those incredible results we like so much.

I recall mention of manufacturer's using A-weighting to get better numbers; is that along the lines of what you're talking about?  Out of curiosity, I looked up the specs of a fair number of DACs last night -- from cheap to expensive -- and have to admit that the results were surprising: the Benchmark (SNR stated as 123 and 126, A-weighted and unweighted, respectively) looked far better than a premium unit that costs $22K (for one channel!).  But even so, does SNR matter?  All the music I have seems to stay far enough away from the noise floor that even SNR specs may be a bit of a red herring in terms of a useful metric.  
 
I've been reading a thread in the hardware forums where user Gary in MD auditioned 14 different DACs and (as of thread page 60/89, which is as far as I've gotten) had a tough time finding any sonic difference between most of the contenders.  People on the sound science side will argue about the validity of his testing methodology, but since it was never intended to be a rigorous scientific evaluation I'm not going to comment.  
 
But...
 
Isn't the fact that he couldn't tell the difference a reasonable result?  The DAC is just supposed to take the 1s and 0s and turn them into an analog waveform, and it's not exactly a digital task that requires bleeding edge technology.  As a mechanical designer, I've had to employ the services of our test and measurement guys every so often, and I've yet to hear one of them say they've taken a new piece of equipment but can't use it until it's "burned in," or that any particular component "colors" the resulting data.  It seems like any reasonable implementation ought to get the job done, and perhaps Gary might have been better served if he's just tried to eliminate the units that sounded "bad."  
 
Which brings me to this question: these days, is there really such as a thing as a "bad" DAC?  I mean, one that absolutely gets things wrong, that doesn't convert the bits to waveforms properly?
 
I won't say my mind's completely made up, but if the amp section reviews well I'm probably just going to get the Oppo HA-1 when it comes out: it's a one-box solution for my music enjoyment needs.  I suppose I'm not a diehard audiophile if I don't want to chase down the last 0.1% of perceived performance from my gear, and I like the fact that the Oppo will play pretty much any format thrown at it (storage is cheap, and I don't want to mess with resampling any high resolution files that might come my way).  Oppo also seems to make good quality components available for a fair price.
 
Doesn't anyone just enjoy listening to their music without being hypercritical about the equipment used?  In the vast majority of cases, the consumer wasn't there when the music was recorded, so who are we to say that our systems are reproducing it accurately?  I'll say it again: I'm glad Gregorio started this thread, because I'll not be buying 24-bit music simply for the sake of a extending a dynamic range that was already adequate at 16/44.1.  
 
Am I missing something?  Is there a flaw in my thinking?  I'm always happy to be educated.

I said that because you were going from 16/24bit to SNR, my point being that usually however good the specs read, any audible noise will be because of the sound system, not because it's "only16bit" and the noise floor is higher than 24bit.
 
 
does SNR matter? I would be tempted to say that it goes as a pack, if the sound system ends up with ungly distortions only 50db under the music, it will probably not matter much that the SNR is only -80db.
 
to me noise values that matters are those I can hear. that's all. I listen to music very very quietly(most of the time) in a very quiet environment on very sensitive IEMs, so what reads on graphs: music @ 0db and noise @ -70db, might end up being me listening @ -30db and noise only 40db under my music if it's not volume dependent.
I've tried to link values from dacs or amp to the actual hiss I can sometimes get, and I can't seem to find a relation between them, so I have concluded that manufacturer specs are often phony, or that those specs didn't tell about something I can hear.
the only thing I could relate to, came from http://www.markuskraus.com/RMAA/rmaa%20complete%20-%20html.html as he makes the measurements from the IEM, it gives the actual idea of what is happening. so those measurements gave me actual usable intels about hiss and showed what I heard on the daps I've owned. not like all noises are actual hiss, but it's the only one that matters to me if it's the only one I can hear ^_^.
sorry if it's a weird answer, but that's all I've got.
 
about DACs, I believe that some are better than others in an audible way. but at the same time, I believe that a bad dap already does an OK job.
a bad headphone is bad, a bad amp can be pretty messy(bad or not with the intended kind of headphones), a bad dac is making ok sound. worst case is usually a smaller soundstage, some kind of roll off due to bad filtering? even with a poor hifimediy I didn't feel like the sound was bad. I preferred the odac or even my daps with a line out, but it wasn't bad. sound differences were audible but not really obvious, and it was more "different" than better or worst. telling who had the truth would have been a lot less obvious if I didn't know one was a cheap little usb crap. I really believe you could trick a few guys in a double blind test with different dacs.
now that's just my opinion based on my still small experience. I've read some stuff that would suggest a lot of ways a dac could go so very slightly wrong. like the fact that the moment the dac is triggered by voltage entering (signal=1) doesn't always start exactly the moment the tension start going through. something about taking the point with maximum voltage as the trigger, and the signal overshooting or undershooting at some places along the duration of the signal 1.  possibly creating some matter of jitter.
 
for most questions, I feel like someone like ClieOS would be better fitted than humble nooby me.

 

[bigshot]

Sounds to me like you've got a piece of equipment in your chain that isn't working properly. None of that makes any sense.

 

[cjl]

Amping isn't as much of an inconvenience for a home or office system I suppose. Worse for a portable system.

True, and the rare times I use a portable systems, I'm using my Senn IE80s, which don't seem to care about source impedance much (so I drive them straight out of my MP3 player without any problem at all). When I'm at home, I tend to use speakers rather than headphones, since they sound so much fuller and more natural than any headphones I've tried. Because of this, the only time I'm left listening to headphones that care at all what is driving them is at work, where having one extra little box sitting on my desk is no big deal (and if anything, I prefer the physical volume knob on the O2 to software volume control, so even from a convenience standpoint, it is an improvement).

 

[castleofargh]

Sounds to me like you've got a piece of equipment in your chain that isn't working properly. None of that makes any sense.

I can't let you say that !!!!!
at least I got that part right

for most questions, I feel like someone like ClieOS would be better fitted than humble nooby me.

 

[bbmiller]

Irrespective of there being any reason for recordings with a 24-bit depth to be better than ones with a 16-bit depth, do you think they are because the producers/audio engineers know when they produce a recording with 24-bit depth they are catering to people who care about sound and thus create better mixes/audio engineering for people who care about better sound?