PMA's listening test files are available for a limited time at: https://www.dropbox.com/scl/fo/mcju...vvj1fR2w?rlkey=xqxccfasytjc0dkfvugzg3njf&dl=0
That particular test used RTZ DAC with PCM2DSD converter and ES9822PRO ADC.What DAC/ADC is that transparent?
This DeltaWave software that is mentioned often here, does it use a DAC of any sorts. If so how can we use it in our comparative test or reference. How did you manage to get the audio files onto the website, are we not just listening to the artifacts from your DAC. Is your replication of the test music perfect to the analogue?
Last edited:
Deltawave works by comparing two audio files. It can correct for level differences between the two files and give you subsample time alignment. You can then compare how closely the files match, and it provides tools for analyzing waveform differences, magnitude differences, phase difference, etc. This does not involve any DAC (or listening).
There is also something called the Deltawave comparator, which is an ABx tester. It takes two audio files and again level matches them, and time aligns them. You can then complete an ABx listening test through whatever DAC you like. In the case be discussed, recordings of the RTZ DAC output using different versions of the PCM2DSD modulator were made with an ADC. There were strong claims about how the recordings sounded different from each other and how they sounded significantly different from the original.
Given the differences reported on listening tests it should be trivially easy to provide ABx results showing the ability to distinguish the original from the recording, but no one has successfully done this. I attempted to do so and failed.
Michael
There is also something called the Deltawave comparator, which is an ABx tester. It takes two audio files and again level matches them, and time aligns them. You can then complete an ABx listening test through whatever DAC you like. In the case be discussed, recordings of the RTZ DAC output using different versions of the PCM2DSD modulator were made with an ADC. There were strong claims about how the recordings sounded different from each other and how they sounded significantly different from the original.
Given the differences reported on listening tests it should be trivially easy to provide ABx results showing the ability to distinguish the original from the recording, but no one has successfully done this. I attempted to do so and failed.
Michael
Well there is always listening to an original file versus one that went thru a DAC to ADC conversion. That takes the issue of the ADC out if that is not detected reliably.The louder one will be preferred, of course. But if it comes out in ABX testing it means people are discriminating one is louder than the other without knowing that's what they are doing. That's because ABX is inherently only for discrimination testing.
For other types of discrimination testing, it depends on what the actual difference is that is being discriminated. Some audible effects are influenced by volume level way more than other things. The example I gave previously had to do with horn player vibrato and whether there was only frequency modulation, or if there was also amplitude modulation too. That particular discrimination task in pretty insensitive to volume level.
Also, if you record the two dacs then try to listen to the two files, if you are using a dac that does not reproduce vibrato amplitude modulation well, then you will probably never hear the difference. Comparisons in DeltaWave may be exactly the same because the average level of vibrato volume level is the same for both dacs (its that one is modulated around the average, and the other isn't). That average is what you see in an FFT analysis. (note: modulation implies sidebands, but its the phase of the sidebands that matters, not just their amplitudes)
The point is that there is no one-size-fits-all solution to trying to figure out if two dacs sound the same or different. That includes whether SINAD is good enough.
I'd still like to find out more about the example you gave of the horn player vibrato. Frankly your description makes no sense to me. You could at least say which two DACs handle this differently. There is more to Deltawave than FFT's. You can listen to the difference file among other things. Or view in detail the waveform of the difference file.
Many of them. I'd offer the RME ADI 2 Pro FS as an example.What DAC/ADC is that transparent?
I think that would be like offering to let a mechanic work on something, but only allowing him to use his pliers instead of his whole toolkit. You seem hung up on FFT issues while wanting not to look at other ways to accomplish things.That could be a good test alright. Have to think about it. I would want to use frequencies that are already used a lot in the music. Maybe change the brush pattern on a snare drum slightly for one beat in the song. Probably I would need all the tracks for a busy recording, so I could mess with just one of them a little bit.
Thinking about what if I swapped the position of two snare brush patterns so they occur at different points in the song, but the song is on average the same.
To make it fair, the DeltaWave operator should not listen to the files so as to be able to focus in on spots that do sound different, only use the FFT analysis to see if it detects the difference unaided.
Just remembered I have some files made by PMA of different opamps. Those are the files I sorted for distortion. I could put them in my dropbox if anyone wants to play with them.
I said it can even be the same dac, just with different clocks.I'd still like to find out more about the example you gave of the horn player vibrato. Frankly your description makes no sense to me. You could at least say which two DACs handle this differently.
Do you want to know what dac it is I used for that? Because its not one you can just go out and buy. You would have to build it. Its not too hard to do if you can solder fine pitch SMD.
It also helps a lot to hear the difference if you have large panel electrostatic speakers, such as Sound Lab. If not that then Quads. Something not as good but still not too bad are Magnepans.
https://www.soundlabspeakers.com/majestic/
Last edited:
I use a DVM, a scope, signal generator, tiny VNA, and various other instruments. However, the one in most people's hands that gets the most overused and most incorrectly interpreted is the FFT spectrum analyzer. Why? Because it can show very small differences, and because people start to believe that whatever they see in that spectrum is all there is that really matters most.You seem hung up on FFT issues while wanting not to look at other ways to accomplish things.
Its why some people say they care more about SQ than measurements. Its because of overfocusing on making the spectrum look pretty can lead to not such good sound. And that's not because people like distortion either. One can have a nice looking FFT and a collapsed sound stage. There is no guarantee the sound stage will be great. Now if the noise skirts that can be viewed are optimized then that may help some. Only one person here might know whether that's the case or not, the guy with the top secret dac, Bohrok2610. And I not sure he would know it if he heard it.
Last edited:
I'm failing to see how that could result simply from a different clock.I said it can even be the same dac, just with different clocks.
Do you want to know what dac it is I used for that? Because its not one you can just go out and buy. You would have to build it. Its not too hard to do if you can solder fine pitch SMD.
It also helps a lot to hear the difference if you have large panel electrostatic speakers, such as Sound Lab. If not that then Quads. Something not as good but still not too bad are Magnepans.
https://www.soundlabspeakers.com/majestic/
I've had Soundlabs for 15 years. Other speakers I've used over the years are Maggies, Acoustats, old and new Quads.
Great! You're in great shape speaker-wise. How is your room?
What dac do you use?
Regarding clocks, you might be surprised how much effect they can have on a good DSD-only dac.
Regarding RME, IME they are a good example of a dac that can measure well yet sound bad. What are they using, an AK4493?
I have an original ADI-2 which I never use because of the bad sound. Also, a forum member who visited my place twice said this after hearing an earlier version of my dac:
"....visiting Mark and hearing his system was eye-opening for me. Since then, I have worked on my DACs and two most important changes that have resulted in sound being closer to natural...My DAC before that was an RME ADI-2 Pro, which has fantastic specs, but sorry to say sounds awful in comparison."
He said a lot more too, but not related to RME. His whole post is at: https://www.diyaudio.com/community/threads/dac-recommendation.376015/post-7560022
What dac do you use?
Regarding clocks, you might be surprised how much effect they can have on a good DSD-only dac.
Regarding RME, IME they are a good example of a dac that can measure well yet sound bad. What are they using, an AK4493?
I have an original ADI-2 which I never use because of the bad sound. Also, a forum member who visited my place twice said this after hearing an earlier version of my dac:
"....visiting Mark and hearing his system was eye-opening for me. Since then, I have worked on my DACs and two most important changes that have resulted in sound being closer to natural...My DAC before that was an RME ADI-2 Pro, which has fantastic specs, but sorry to say sounds awful in comparison."
He said a lot more too, but not related to RME. His whole post is at: https://www.diyaudio.com/community/threads/dac-recommendation.376015/post-7560022
Last edited:
I have several DACs I can use. RME Babyface, Antelope Audio, Focusrite, Topping and some others. I mostly concur with bohrok2610's comments. Such descriptions from a casual listening session don't mean very much. The conclusions offered were overwrought.
I don't use DSD though some of my DACs support it.
I don't use DSD though some of my DACs support it.
This DSD dac has bult-in conversion of PCM to DSD256. So it can player either format.
All your ESS, AKM, etc., type dacs are remodulating the audio too, but maybe to 5-bits instead 1-bit.
1-bit is inherently more linear, yet also more sensitive to jitter. There are other issues between the architectures as well.
Regarding your dac collection, it seems like they might be a bunch of mid-tier consumer market dacs. That's a common path people take as the keep believing in measurements. I went through a lot of that too. Only I built some of my own dacs. Found out that ES9038Q2M sounded better if fed DSD from an AK4137 ASRC chip. Why? I didn't know, but I stuck with it. Ended up with a dac that was pretty hard to tell apart from Benchmark DAC-3 by ear. It was only later I learned what reproduced audio is really supposed to sound like, which a great stereo illusion of sound stage. It is apart from steady state distortion and noise measurements. Your particular speakers won't reproduce sound stage well unless the room, the speaker placement, dac, etc., are all correctly setup and working. Otherwise the speakers will tell you the truth that you don't have it right.
All your ESS, AKM, etc., type dacs are remodulating the audio too, but maybe to 5-bits instead 1-bit.
1-bit is inherently more linear, yet also more sensitive to jitter. There are other issues between the architectures as well.
Regarding your dac collection, it seems like they might be a bunch of mid-tier consumer market dacs. That's a common path people take as the keep believing in measurements. I went through a lot of that too. Only I built some of my own dacs. Found out that ES9038Q2M sounded better if fed DSD from an AK4137 ASRC chip. Why? I didn't know, but I stuck with it. Ended up with a dac that was pretty hard to tell apart from Benchmark DAC-3 by ear. It was only later I learned what reproduced audio is really supposed to sound like, which a great stereo illusion of sound stage. It is apart from steady state distortion and noise measurements. Your particular speakers won't reproduce sound stage well unless the room, the speaker placement, dac, etc., are all correctly setup and working. Otherwise the speakers will tell you the truth that you don't have it right.
Last edited:
You have do listening tests, there is no way around it. Its can be useful statistical evidence. JBL uses trained listeners to evaluate their speakers. In their experience and in mine, training listeners makes a big difference in their reliability.
Besides, its not just a voltage we are trying to regulate. Its also the room acoustics as the human hear hears them. I used to have video display between my speakers. That had go, because of acoustical reflections. Everything matters if you want to get to the truth of what is actually encoded on a CD.
Regarding noise, distortion, FR, type metrics, I don't think those are the only ones. If you have followed this thread you should know there are dynamic noise effects in dacs that do not show up in typical spectral analysis. Those things are well known to exist by expert dac designers, but they are hard to measure. Maybe there is some progress towards a new way of measuring, but it still requires very expensive test gear.
I don't know if you will be familiar with some of the terminology in the attached paper, but it might be worth a look if you are.
Besides, its not just a voltage we are trying to regulate. Its also the room acoustics as the human hear hears them. I used to have video display between my speakers. That had go, because of acoustical reflections. Everything matters if you want to get to the truth of what is actually encoded on a CD.
Regarding noise, distortion, FR, type metrics, I don't think those are the only ones. If you have followed this thread you should know there are dynamic noise effects in dacs that do not show up in typical spectral analysis. Those things are well known to exist by expert dac designers, but they are hard to measure. Maybe there is some progress towards a new way of measuring, but it still requires very expensive test gear.
I don't know if you will be familiar with some of the terminology in the attached paper, but it might be worth a look if you are.
Attachments
Last edited:
Ordinary dacs, both better than -120db for thd+ noise and imd 32 tone. So competent.Were they just ordinary DACs, or especially good ones?
I assume for the purposes of your test that the rest of the system was very good?
We seem to have stories going both ways so I’m just interested to find out more.
I can’t resist: If 2 DACs output the same waveform at the same base level, do they sound the same?
Measured how, with an oscilloscope?
What is the output impedance of each dac, the same for both?
Of course if everything is identical down to the most microscopic detail, how could they sound different?
What is the output impedance of each dac, the same for both?
Of course if everything is identical down to the most microscopic detail, how could they sound different?
An easy question, without an easy answer. The issue comes down to how you define them as having the same waveform. As seen from an oscilloscope, you can’t accurately assess whether the waveforms are truly the same. It’s possible to compare signals via a differential amplifier, and measure the residual noise and distortion at the output, but even that has faults and limitations.I can’t resist: If 2 DACs output the same waveform at the same base level, do they sound the same?
- Status
- Not open for further replies.
- Home
- Source & Line
- Digital Source
- can DACs sound different if they both measure well?