That depends on what the goal of the audio equipment is. If the goal is to reproduce what is on the recording as closely as possible basic engineering practice is to measure e.g. frequency response, distortion, noise, crosstalk. With DACs typically also linearity and jitter.
If the goal is to produce subjectively pleasing sound then you first need to define what is a pleasing sound to you. Then you can take measurements to see how they differ from another device that does not sound subjectively pleasing to you.
It goes with SINAD, very simple measurements and nubmers... Nothing special, nowadays you can put your finger somewhere in left side of one picture from ASR table with results and get a brilliant by measurements cheap device (will you buy it?), but it will not sound as yours DIY DAC with the same chip, because of whole device implementaion, layout, power supply, and each component, etc. So the main difference in sound color/portrait is not in these measurements, all modern DACs are good enough to pass cool hi-fi numbers in measurements but this is not about how it sounds.
Last edited:
No, SINAD as a number does not tell much and ASR (or AP) is not all there is to measurements. HD profile, noise (including out-of-band noise), close-in PM/AM noise can show differences between high SINAD devices. This fixation with ASR SINAD numbers seems to plague every discussion of measurements here. To state "Anything that is audible to humans can be measured" does not mean that ASR measurements, especially SINAD, are such measurements.
not if my opinion is that blind tests are inherently flawed.... i already did a few blind tests in the digital domain in terms of ABX software... and DBT has its purpose but its not the holy grail of test methods, i would never have learned how absolute polarity sounds like if i only did blind tests for examples and the general "science consensus" on absolute phase is that its inaudible, its not....
well the thing is even if we have measurements showing X reduction at Y frequency, the thing that would matter the most (atleast to me) is how it sounds in the end, and i dont need measurements to know that this filter "does something"
i delibaretly choose the last ferrite bead to be effective even in the 2-3Ghz range (WLAN) while offering low DC resistance (22mOhm)
add to this the pi filter configuration
and the common mode choke on usb lines:
very good to filter out usb signal harmonics from what i understand , and its even a bit effective at wlan range (around -15db)
from what i read thats not only it, just part of it..... the general high impedance will also REFLECT noise, searching some more low impedance path, thats where its crucial to not leave a "open drain" to your hifi system
Why don't we start with something every human has some capacity to hear, sound stage. How do you measure sound stage (width, depth, and precise localization of different instruments playing at once)?
Actually, at least to some extent some substantial work has already been done in this area, as full spatial pick-up microphones now exist (displacement and pressure in three orthogonal directions). However, not everyone has the equipment, including the room, to acquire the data to be analyzed with a complex spatial microphone. Then comes the problem of analyzing the data. FFTs are likely involved, but spectral analysis by itself would be far from sufficient.
If we initially limited the problem to 2-dimensional (width, depth, and localization precision), and of we left out speakers and rooms, we already have electronics with varying capacity to reproduce sound stage, including localization cues. The simplified problem would still remain as to mathematical analysis of a dataset.
Actually, at least to some extent some substantial work has already been done in this area, as full spatial pick-up microphones now exist (displacement and pressure in three orthogonal directions). However, not everyone has the equipment, including the room, to acquire the data to be analyzed with a complex spatial microphone. Then comes the problem of analyzing the data. FFTs are likely involved, but spectral analysis by itself would be far from sufficient.
If we initially limited the problem to 2-dimensional (width, depth, and localization precision), and of we left out speakers and rooms, we already have electronics with varying capacity to reproduce sound stage, including localization cues. The simplified problem would still remain as to mathematical analysis of a dataset.
Last edited:
I must applaud the above statement. Thank you for saying it.
Blind tests are the cornerstone of all scientific research. Your opinion may be that blind tests are flawed because they do not back up your subjective findings....
Also can you stop quoting more than one person in a reply, please
Last edited:
i simply dont believe science has discovered everything yet, maybe due to flawed testing methods or/and the overall nature of (subjective) audio listening
science found a way to eliminate bias, but that doesnt mean that this is the ultimative testing method for "everything", feel free to believe so yourself tho
No lol, should i do 3 posts instead?