Indeed. Leaving aside whether the measurements were well done (not something one can necessarily assume), they can fail to present an accurate portrait and comparisons for two reasons: 1) the measured differences are inaudible, for instance well below the noise floor (don't laugh), and 2) the measurements are time/frequency averages that miss subtler aspects of sound dynamics that matter in actual perception. Standard measurements can detect gross weaknesses, but once they look good within the audible range, they provide no additional information for selecting a DAC.
I reckon with #2 you've hit upon an important point missed in most arguments about audibility. There are, in my mind, two levels of "audible". The first is what I call "overt" audibility, that is signals above -95dB which it is possible to make out clear differences, to varying degrees. The second deals with what you mention: Aspects of the signal that our brain interprets spacial and other information from that is below what we can readily discern in, say, short A/B tests.
I reckon with #2 you've hit upon an important point missed in most arguments about audibility. There are, in my mind, two levels of "audible". The first is what I call "overt" audibility, that is signals above -95dB which it is possible to make out clear differences, to varying degrees. The second deals with what you mention: Aspects of the signal that our brain interprets spacial and other information from that is below what we can readily discern in, say, short A/B tests.
let's say we open the door to consider tiny variations with no immediate conscious impact under controlled listening so we fail the test. let's say we agree(despite all that being basically born out of sighted impressions and made up ideas instead of actual data) that such tiny variations may possibly affect us in a deeper subconscious way, or maybe have some cumulative impact that ends up creating a conscious change after a while. we then have to be honest and fully open that door. we have to accept that pretty much everything in life satisfies the criteria. the smell of my shampoo, my lack of magnesium, memories of my first crush as a kid, if my volume level is set 0.03dB louder, if a car is passing 5km away from here... billions of variables with most of them making us who we are, how we feel, and how we interpret things at every moment of our lives. also a great many variable having pretty much no impact whatsoever despite being perceived, and even more being simply too small to be registered by our sensors.
there is no evidence that some signal at -95dB has more impact on my experience of an album than most other stuff including where I place the lamp in the room, what I had for lunch, my mood, and the temperature being 2° hotter. in fact if I was to rank their impacts over my "listening" session based on my understanding of humans, I'd be strongly tempted to put all those variables above a signal at -95dB. anything below JND(just noticeable difference) in a controlled test has no legitimate reasons to be seen as more than a drop in the ocean of potentially significant variables we never cared about because there are just too many of those. and certainly such potentially significant variable below JND is unlikely to provide evidence of being as significant as most of the already noticeable differences, audio or not. so it's like a drop in the ocean while we're looking at some cloud in the sky. IMO only wishful thinking supports the idea of significance without demonstrated significance.
I reckon with #2 you've hit upon an important point missed in most arguments about audibility. There are, in my mind, two levels of "audible". The first is what I call "overt" audibility, that is signals above -95dB which it is possible to make out clear differences, to varying degrees. The second deals with what you mention: Aspects of the signal that our brain interprets spacial and other information from that is below what we can readily discern in, say, short A/B tests.
My point #2 was not about audibility in human subject experiments, but rather that measurements based on time-frequency averages cannot represent faithfully all that is already known about human hearing. That's why even the best hearing and signal processing researchers I knew at Bell Labs did many psychophysic experiments in addition to system measurements in inventing perceptual coding, steerable microphone arrays, and the like. And that's why we need to complement objective measurements with actual audition.
This is an excellent response . The Yggy has something that sounds closer to live music than any delta sigma dac I’ve listened to . I had a Chord 2Qute (not delta sigma) that I’m sure measured better and it’s a fine dac but the Yggy wiped the floor with it . I sold it and bought another Yggy .
Listening to the plucking of an acoustic guitar on the Yggy is so realistic , you would think the band is playing right in your room .
That’s all I ask of any dac and I will never sell it because of this
Sorry to necro on an old thread but I must build on this.
I currenly have a Bifrost Multibit (Bimby) and tried a Chord 2qute at a point in time. I couldn't stand it for more than 10 minutes, the only thing it did better was bass.
Sure it sounded cleaner (hence the THD+N measurements it produces), but it just didn't sound like real music, instruments and voices had a synthetic feature that is very hard to describe.
So indeed multibit might measure worse but delta sigma doesn't sound like real music.
FYI, the 2Qute is PWM and not Delta Sigma. Also, not all Delta Sigma sounds terrible and not all R2R sounds great. Of course I get your point with measurements and enjoyment is the main factor. I’m not a fan of over-generalizing.
If a DAC is using PWM then it's got to have a Delta Sigma modulator in there somewhere. Although its theoretically possible to do PWM in NOS form (needs sub-nanosecond resolution on the pulsewidths i.e. an effective clock rate >1GHz) I don't believe anyone's done it yet. Happy to be proved wrong though.
If a DAC is using PWM then it's got to have a Delta Sigma modulator in there somewhere. Although its theoretically possible to do PWM in NOS form (needs sub-nanosecond resolution on the pulsewidths i.e. an effective clock rate >1GHz) I don't believe anyone's done it yet. Happy to be proved wrong though.
To be fair, it’s not exactly PWM and the filter may be more similar to DS. Again, it’s different. The following link is a discussion about how the Pulse Array DAC requires PCM to work and all DSD is converted to PCM with his DACs, unlike DS:
Thanks, those help and they confirm that Rob Watts is indeed using S-D (he didn't say it explicitly but he talks of his 'noise shaper' and says the output from that is 6bit chunks).
However just because he uses S-D doesn't make his DAC equivalent to an S-D DAC, his analog stage is way, way different to how ESS implements it. He's using time slicing to implement his bit weightings, ESS uses on-chip resistors. Hence he has no need for any DEM (dynamic element matching) because slicing in time is inherently linear.
I do agree incidentally with your earlier 'not all D-S sounds terrible' as I've heard PWM-based DACs and they don't have the characteristics I had previously (and erroneously) attributed to the use of S-D techniques.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
