Which DAC do you prefer?

[esldude]

Perhaps the differences are nowhere near as obvious after you download and play them, as they were to Dennis when he originally uploaded them ? Even the choice of player software matters too.

Has Dennis downloaded his own files and compared them with the original .wav files by careful listening ? Surely he wouldn't have bothered uploading them if the original differences were vanishingly small ?

Yeah, I know, "Bits are Bits", and nothing else matters !

 

I have indeed downloaded the same files you have access to. I can indeed hear the same differences in most of them, and also there are a couple I can't parse out (and couldn't before the upload). Bits are indeed bits, and none of these files have the identical bits.

 

Is it so hard guys to download, listen, and pick a preference? If you think they all sound the same, then say so. We can discuss why that might have been the result. At least have the wherewithal to say that is your experience upon listening. Can you simply never bring yourself to say, "well they sound the same to me". Is that not a possibility? Wow if so that is in itself interesting.

 

So listen, be honest, and report what you hear. That is all I ask.

[esldude]

Yeah, I know, "Bits are Bits", and nothing else matters !

 

Have you downloaded the files and listened to them Alex? If you have what did you think?

 

If you haven't, then why not?

[sandyk]

Have you downloaded the files and listened to them Alex? If you have what did you think?

 

If you haven't, then why not?

 

 

 

Simply because I don't feel the need to do so due to the huge numbers of variables involved, unlike the files that Mani posted in the Schiit thread which started out as 24/96, and were only downsampled to 48K,although I would have preferred the original 24/96 to make the differences more obvious. Besides which, there were only a total of 4 files involved in that case.

Neither am I familiar with the majority of selections available, in order to be in a position to judge which version is truer to the original recording, and not just a personal preference.

With Mani's files it was a clear choice to me of choosing which file was degraded, and which file sounded more correct.

[esldude]

Simply because I don't feel the need to do so due to the huge numbers of variables involved, unlike the files that Mani posted in the Schiit thread which started out as 24/96, and were only downsampled to 48K,although I would have preferred the original 24/96 to make the differences more obvious. Besides which, there were only a total of 4 files involved in that case.

Neither am I familiar with the majority of selections available, in order to be in a position to judge which version is truer to the original recording, and not just a personal preference.

With Mani's files it was a clear choice to me of choosing which file was degraded, and which file sounded more correct.

That's a pathetic excuse Alex.

[sandyk]

That's a pathetic excuse Alex.

 

As far as I am concerned there are far too many variables to make the exercise worthwhile, and I am not the only one to suggest this. Give me the simple choice of trying to choose between what you genuinely believe are your best and worst sounding DACs in a PM/email and I will have a listen to a couple of musical selections.

[Sal1950]

Seems like some are just scared. LOL

 

I downloaded the Terrel, McClain, Davis, Mayall, and Hiatt files.

Listened to on my Senn HD650 cans plugged directly into the Emotiva DC-1 DAC.

At this point I've not voted and do temporarily fly the white flag.

A number of times I thought for sure I was hearing a difference in a few files but when I'd jump back and forth between them the difference would disappear.

I'll give it another go in the morning with relaxed ears and see if I can do anything of note.

Thanks for the effort Dennis!

[Sal1950]

As far as I am concerned there are far too many variables to make the exercise worthwhile, and I am not the only one to suggest this. Give me the simple choice of trying to choose between what you genuinely believe are your best and worst sounding DACs in a PM/email and I will have a listen to a couple of musical selections.

So you want to hold your testing to be private, what are you afraid someone will see Alex?

[esldude]

As far as I am concerned there are far too many variables to make the exercise worthwhile, and I am not the only one to suggest this. Give me the simple choice of trying to choose between what you genuinely believe are your best and worst sounding DACs in a PM/email and I will have a listen to a couple of musical selections.

 

So I give you two rather than five choices and you can do it, but five is too much for you? Again pathetic. I wouldn't even have imagined it.

 

I'll discuss the variables if you will give it try like it is. Otherwise it smacks of a total lack of confidence.

[sandyk]

So you want to hold your testing to be private, what are you afraid someone will see Alex?

 

Believe what you want.

I feel that it is a pointless exercise in it's present form, and that there would have been many more participants if he had done what I suggested , unlike that from Mani in the Schiit thread where I did participate. If you want to know whether my posted results were correct or not, i.e. as Mani heard them, you will need to ask Mani ,as he hasn't actually posted which was which after the exercise, unless I missed it.

[esldude]

Believe what you want.

I feel that it is a pointless exercise in it's present form, and that there would have been many more participants if he had done what I suggested , unlike that from Mani in the Schiit thread where I did participate. If you want to know whether my posted results were correct or not, i.e. as Mani heard them, you will need to ask Mani ,as he hasn't actually posted which was which after the exercise, unless I missed it.

 

So what are you saying Alex? Any redbook CD is perfectly reproduced by any decent DAC, but not higher rez material? I still don't get your problem with this. Again, listen, give us a report. You hear it the same say so. We can discuss why that might be. If you don't participate, then what is there to proceed with?

[sandyk]

So I give you two rather than five choices and you can do it, but five is too much for you? Again pathetic. I wouldn't even have imagined it.

 

I'll discuss the variables if you will give it try like it is. Otherwise it smacks of a total lack of confidence.

 

No, it's more like a total lack of interest.

Even with the CD-R that I sent you, you only have 8 pairs of tracks, and have only to decide whether the tracks sound the same, or whether you have a repeated preference for the same one in each pair.

[esldude]

Well, maybe it is too early, but there seems little participation in this listening project so far. Two votes and one person saying they didn't vote.

 

So I am going to reveal the gear used though not which file it matches. That still doesn't tell you which you would prefer, and there is no wrong preference. I wondered if people would as a group cluster around the more accurate DACs or not. If they clustered for the worst as their preference that was okay.

 

All right one of the DACs isn't. It is the original file upsampled to 48 khz. Something of a hidden reference.

 

The best measuring DAC is a Focusrite Scarlett 18i20 recording interface. IMD twin tones results show -113 db at 1 khz and two spikes at -107 db at one khz above and below the tones. A good not great noise floor of -107 db. Very flat frequency response.

 

Subjectively the Scarlett sounds pretty transparent to source. I would say it adds a bit of artificial heft to the low end. The kind of thing that might make a component sound a bit sluggish though in this case it has a bit of heft while staying fleet of foot.

 

The next best measuring DAC I used was a 15 year old Tact RCS 2.0 in bypass mode of course. Response is flat other than drooping about a half decibel at20hz and 20 khz. IMD twin tones show a 1 khz result at -107 db. Noise floor is a about the same as the Scarlett at -107 db.

 

Subjectively the Tact seems to add a bit of space, air and delicacy to the sound. A very audiophile-like trait. It is a slight difference. I am not even sure it is real. Perhaps the difference is the Scarlett has a black faceplate and the Tact a silver one.

 

Next we have the sound card in my 6 year old Lenovo T410 laptop. It actually is relatively competent. Response droops just a bit at frequency extremes. The noise floor is only - 85 db or so. Meaning maybe 14 bit performance. Distortion level in a twin tone IMD test shows a 1khz result of -100 db.

 

Subjectively the Lenovo played over a big system is not bad. You wouldn't comment much other than I think it has a mechanical quality. Nothing you can pick out, but music doesn't seem to flow with it. Then again I know in my mind it is part of a mechanical device.

 

Finally I used the headphone out of my Nexus 6P phone. Response droops a good bit at the frequency extremes. The noise floor is only 75 db down. So maybe 12 or 13 bit performance. IMD distortion is around -90 db at 1 khz and there are a few other sum and differences tones at that level.

 

Subjectively the Nexus has a soft, gentle sound that some people might call a natural sound. Probably a good thing for something that might be used with cheap ugly sounding earbuds.

 

So there you are. Try the files and see what you think. Pick a best and worst. That may or may not be the best and worse measuring.

[Paul R]

Well, maybe it is too early, but there seems little participation in this listening project so far. Two votes and one person saying they didn't vote.

 

So I am going to reveal the gear used though not which file it matches. That still doesn't tell you which you would prefer, and there is no wrong preference. I wondered if people would as a group cluster around the more accurate DACs or not. If they clustered for the worst as their preference that was okay.

 

All right one of the DACs isn't. It is the original file upsampled to 48 khz. Something of a hidden reference.

 

The best measuring DAC is a Focusrite Scarlett 18i20 recording interface. IMD twin tones results show -113 db at 1 khz and two spikes at -107 db at one khz above and below the tones. A good not great noise floor of -107 db. Very flat frequency response.

 

Subjectively the Scarlett sounds pretty transparent to source. I would say it adds a bit of artificial heft to the low end. The kind of thing that might make a component sound a bit sluggish though in this case it has a bit of heft while staying fleet of foot.

 

The next best measuring DAC I used was a 15 year old Tact RCS 2.0 in bypass mode of course. Response is flat other than drooping about a half decibel at20hz and 20 khz. IMD twin tones show a 1 khz result at -107 db. Noise floor is a about the same as the Scarlett at -107 db.

 

Subjectively the Tact seems to add a bit of space, air and delicacy to the sound. A very audiophile-like trait. It is a slight difference. I am not even sure it is real. Perhaps the difference is the Scarlett has a black faceplate and the Tact a silver one.

 

Next we have the sound card in my 6 year old Lenovo T410 laptop. It actually is relatively competent. Response droops just a bit at frequency extremes. The noise floor is only - 85 db or so. Meaning maybe 14 bit performance. Distortion level in a twin tone IMD test shows a 1khz result of -100 db.

 

Subjectively the Lenovo played over a big system is not bad. You wouldn't comment much other than I think it has a mechanical quality. Nothing you can pick out, but music doesn't seem to flow with it. Then again I know in my mind it is part of a mechanical device.

 

Finally I used the headphone out of my Nexus 6P phone. Response droops a good bit at the frequency extremes. The noise floor is only 75 db down. So maybe 12 or 13 bit performance. IMD distortion is around -90 db at 1 khz and there are a few other sum and differences tones at that level.

 

Subjectively the Nexus has a soft, gentle sound that some people might call a natural sound. Probably a good thing for something that might be used with cheap ugly sounding earbuds.

 

So there you are. Try the files and see what you think. Pick a best and worst. That may or may not be the best and worse measuring.

 

 

Eh, it more just a lack of time. I plan to download and listen to them tonight or tomorrow.

 

I think this is a very well thought out test, and just ignore the folks howling about you not doing it *their* way. They obviously won't spend the time necessary anyway, unless it is guaranteed to prove one of their pet theories.

 

-Paul

[maelob]

Seems like some are just scared. LOL

 

I downloaded the Terrel, McClain, Davis, Mayall, and Hiatt files.

Listened to on my Senn HD650 cans plugged directly into the Emotiva DC-1 DAC.

At this point I've not voted and do temporarily fly the white flag.

A number of times I thought for sure I was hearing a difference in a few files but when I'd jump back and forth between them the difference would disappear.

I'll give it another go in the morning with relaxed ears and see if I can do anything of note.

Thanks for the effort Dennis!

I feel so much better, we are in the same boat!

[manisandher]

All right one of the DACs isn't. It is the original file upsampled to 48 khz. Something of a hidden reference.

 

If these are the same files as used in the ADC test thread, then I think there is an issue. To me, there is something clearly amiss with the way they sound - thin, 'sizzly', and generally anaemic. I strongly suspect their character has been changed in the upsampling process.

 

What this means is that my subjective preference will likely be for a DAC that 'tames' the ref files the most. And I don't think this is very helpful.

 

Mani.

[esldude]

If these are the same files as used in the ADC test thread, then I think there is an issue. To me, there is something clearly amiss with the way they sound - thin, 'sizzly', and generally anaemic. I strongly suspect their character has been changed in the upsampling process.

 

What this means is that my subjective preference will likely be for a DAC that 'tames' the ref files the most. And I don't think this is very helpful.

 

Mani.

 

I don't find the sound of the upsampled file to be different than the original. Which doesn't mean someone else couldn't. I do have this.

 

 

This is the FFT of having taken a musical file, upsampled to 48 khz and downsampled to 44.1 khz. The noise at the high end is simply from dither used when resampling. For the 48 khz version there would be less of this. The entire band below 20 khz would amount to slight noise at around -140 db as the difference in the two versions. I used TDPF dither for this resampling.

 

I can run some test tones and such through the resampler and record that versus the original and see what differences there are. Will have to wait to later as I will be away from home most of the next few days.

 

It would be nice if we had plenty of participants who could perhaps tell us if they are hearing the same thing as you are. Unfortunately, not the case as of right now.

[manisandher]

I don't find the sound of the upsampled file to be different than the original. Which doesn't mean someone else couldn't.

 

Maybe there's something the re-sampler is doing that I'm particularly sensitive to? Many years ago I tried a few software re-samplers and hated all of them. I did a few resamples using the functionality in the PM2 - the results were much better than any of the software re-samplers, but still not great. Why this should be is totally beyond me.

 

I can run some test tones and such through the resampler and record that versus the original and see what differences there are. Will have to wait to later as I will be away from home most of the next few days.

 

That'd be interesting...

 

Mani.

[esldude]

Maybe there's something the re-sampler is doing that I'm particularly sensitive to? Many years ago I tried a few software re-samplers and hated all of them. I did a few resamples using the functionality in the PM2 - the results were much better than any of the software re-samplers, but still not great. Why this should be is totally beyond me.

 

 

 

That'd be interesting...

 

Mani.

 

Okay, got a chance to do the test tones on this. I sent the original 44.1 khz test signals through a DAC to the ADC and immediately afterwards sent the same signal which was upsampled to 48 khz thru. Both signals were recorded at 48 khz by the ADC. Also I used slightly different windowing on the FFT so the numbers vary slightly though not by any important amount from my earlier numbers.

 

On the 1khz tone at - 1db>

For the original file the second harmonic lies at -118.4 and the third at -104 db with the 4th in the noise floor and the fifth at -130 db.

For the resampled file the second harmonic lies at -119.4 and the third at -104.6 db with the 4th in the noise floor and the fifth at -127.2

 

IMD twin tone:

Original 1 khz is at -114.4 db. 1khz above and below the test tones is -92.8 and -93.8 db respectively.

Resampled 1 khz is at -116.8 db. 1 khz above and below the test tones is -92.8 and -93.7 db respectively.

 

There are a couple other tones at lower levels on both.

 

The J-test signal for both were good and the first few spurs differed by a 1db or less between the two versions.

 

Frequency response was the same other than the dither noise above 22 khz in the resampled 48 khz file.

 

Noise floors while playing a -60 db tone were identical except the resampled file showed a roughly 6 db increase below 8 hz.

 

A test tone I have seen trip up many resamplers is a dual sweep. A 2hz-20,000 hz sweep concurrent with a 20,000hz - 2hz sweep. Some show all sorts of garbage in the spectrogram view. Below are the recorded results for the original 44.1 file and 48 khz resampled file recorded.

 

The background being gray indicates no signals above - 120 db (using a 4096 pt FFT). The red X is the dual sweep. The lighter blue lines are sum and difference tones. In this case most of the blue would indicate a level around or just below -100 db.

 

 

There would appear to be no reason the resampler (Sox in this case) would be audible. If anything the resampled file may be a bit better measuring. Maybe I should try this with 44.1 originals and the resampling to 96 and 192 rates to see if measures improve some more.

 

My reasoning for having a resampled file was so one of the files was the original. However the original being a different sample rate would have been obvious. Resampled to 48 khz it wouldn't be noticed and in my experience would sound the same as the original effectively accomplishing what I wished for this.

 

Now I could tell everyone what the resampled file is so it could be ignored, but I am not much inclined to do so with close to no participation in the poll thus far.

[manisandher]

Okay, got a chance to do the test tones on this.

 

Thanks for looking into this Dennis. Unfortunately it doesn't help in my case - there's still something about the resampled files (assuming I've identified them correctly - pretty sure I have) that slightly grates on me. But in any event, they're still clearly the best sounding files out of the 5 - the most dynamic, most solid, and cleanest sounding. So I've voted them as my preference. Weirdly, the worst sounding files share a similar tonality to the upsampled files, but absolutely none of the good traits.

 

Thanks for putting the test together. Very interested in knowing the 'correct result'.

 

Mani.

[manisandher]

Seeing as I've completed both tests (ADC and DAC), Dennis has been kind enough to share the 'truth' with me. Of course, I won't disclose anything to anyone, but the key takeaway for me is that the most accurate digital files are those that sound the most analogue. Surprise, surprise! Don't be seduced by artificial detail and false transparency.

 

Mani.

[MrSprout]

If I had to guess I'd go with the following:

 

A - Upsampled file

B - Nexus

C - Focusrite

D - Lenovo

E - Tact

 

A and C are virtually indistinguishable to me but C seemed to have a tiny fraction more bass, so based on the subjective comments I'll say that's the Focusrite. E was most transparent, so that's the Tact. B was a bit different and rounded, so that's the phone, and D was boring, hence the Lenovo.

 

If I've got them wrong I can just blame my cheap DAC. If I've got them right I'll also blame my cheap DAC for preferring the mobile phone!

[manisandher]

If I've got them wrong I can just blame my cheap DAC. If I've got them right I'll also blame my cheap DAC for preferring the mobile phone!

 

Now that's the spirit! For anyone else following this thread, just see this as a bit of fun... with perhaps an interesting outcome.

 

Mani.

[esldude]

Of the more than 2000 who viewed this thread, can a few of you explain why you didn't bother to vote in the poll or respond with a post? As Mani said, and with responses like MrSprout it could have been a very fun interesting discussion. This would in some ways have been a bit more trouble and effort than the other listening polls I put up recently, but also could have been the most fun.

 

So was this simply of no interest at all? Or what was holding people back from participating?

[Sal1950]

I did reexamine the files a few times as I promised I would Dennis. I never did feel comfortable enough in my results to actually make any poll entries. I didn't take your suggestion to maelob to "select all of them as preferred and sounding worst" seriously and felt it would just confuse the tabulations.

[Rock Bottom]

Again with the listening to recordings of equipment...I recall folks were listening to recordings of cable, now we're listening to recordings of DACs?

 

And...someone slyly questions if such a fuss is what gives audiophiles a bad name.

 

Fear not; there's nothing remotely audiophile about this post.