General debate thread about audio measurements

[pkane]

@Blumlein 88
Tests i ask are very clear:
Why not making same test with 768khz samplerate to see if it's good or not to convert to this format.

No test do that then no test validate that reading high res on a dac performs good.

I don't ask to be convinced i just ask tests to validate things that are not validated.

Personally, I wouldn't be happy with any test that doesn't measure into many hundreds of gigahertz. Why stop at 768k? What if there's a large spike at 390kHz and you miss it? Better be safe than sorry. I wouldn't want to get a sub-par DAC that doesn't perform well at 500 GHz, for example, would you?

 

[mansr]

Personally, I wouldn't be happy with any test that doesn't measure into many hundreds of gigahertz. Why stop at 768k? What if there's a large spike at 390kHz and you miss it? Better be safe than sorry. I wouldn't want to get a sub-par DAC that doesn't perform well at 500 GHz, for example, would you?

My spectrum analyser only goes to 3 GHz, but this is what I get from an iFi Nano DAC:

 

[pkane]

My spectrum analyser only goes to 3 GHz, but this is what I get from an iFi Nano DAC:
View attachment 28868

I still have to wonder what might hiding just off to the right, there....

 

[mansr]

I still have to wonder what might hiding just off to the right, there....

Well, there's another wifi band at 5 GHz.

 

[LTig]

Well, there's another wifi band at 5 GHz.

Going further to the right we finally reach the optical spectrum. Surely a serious scientific cause against blind testing of audio components

 

[Calexico]

Also switching psu can put ultrasonic garbage. isn't it good to show it?

 

[Duke650]

 

[777]

Maybe if the ultrasonic images will be push up in a higher frequencies we will have a a simpler analogue filter so, the phase will be more benign ?

 

[solderdude]

That's what oversampling does.
Then again to properly oversample you need a steep filter in there to prevent aliasing.

 

[Bds3151]

I'm continuing a discussion from another thread...

See...
22 minutes ago
#223
amirm said:
How did you go from what I said to this? I said that 116 dB is the provable inaudibility across all content and all people. That doesn't mean that *I* can grab some piece of content and demonstrate it with my ears for all of you, and content. It is like me saying you can't get sick if you have pure water and you asking me that if you had one microbe to it, it will make you sick. It may, or may not. There is doubt there whereas there is no doubt with pure water.

The dynamic range of human hearing is calculated both using listening tests and limits of noise floor in our hearing system.

Anyway, none of this is specific to the DAC in this review so follow ups should be to this thread: https://www.audiosciencereview.com/...-debate-thread-about-audio-measurements.2580/
Ok I'll move my comments. Amirm, I didn't say you could, I asked if you could differentiate or develop some differentiation at another level. It was a question.


So there is information on the web re: tests for audibility of distortion. This particular research indicates that it has to reach 1 to 3%. Is it possible to verify this and integrate it somehow into the results?
https://www.axiomaudio.com/blog/distortion/

 

[RayDunzl]

So there is information on the web re: tests for audibility of distortion. This particular research indicates that it has to reach 1 to 3%. Is it possible to verify this and integrate it somehow into the results?

1% thd for a single harmonic = -40dB relative to the level of the fundamental.

3% = -30,45dB

 

[Blumlein 88]

I'm not too surprised by the effects though maybe a little. One reason I don't like SINAD is it combines noise and harmonic distortion. I think basically with HD, it has been shown with test signals audibility disappears below .1% or -60 db. With music the number has to be higher though one can always find some weird music that might be like a test signal. I wouldn't be suprised if one is hard pressed to find any music that less than 1% would be audible. At normal listening levels simple noise added is audible at lower levels.

I understand what @Bds3151 is getting at here. The problem is such guidelines are going to be drawn with very fuzzy margins. Sharper margins like SINAD below -116 db are surer. Or something like a system should at least meet full 16 bit performance for noise and distortion. The odds are if you manage distortion below -70 db (just for a margin of safety) and the gear is quiet enough you don't hear noise at any level (-100 db would do for almost any purpose), and it is flat response +/- .05 db 20 hz to 20khz you'll be golden with anything. But you can't categorically say it is 100% surefire golden. Surprisingly one of the most difficult to get with these relaxed metrics is going to be flat frequency response. Also you do need to take into account gain staging or you'll lose some of these qualities between source and speak input.

I do think Amir's testing has more value than perhaps the above paragraph might imply. I think the next places to be useful in a more cogent categories are amps with real loads attached, and speakers. He obviously wants to move in those directions, but they are harder to manage.