I ran it 10 times on closed-back headphones with active noise cancelling and within a few hertz it concluded that I am 19. Did the same on open-back headphones I was 36 which was probably due to outside noise leakage. I always though that the noise cancelling was sort of gimmickry but now I am confused. 😕
That fairy story you just told is an hallucination of Amir. It assumes noise floor is the the only noise effect in a dac, and that "distortion" (i.e. spurious spectral lines) is the only other defect in sound that can exist. Its surprising how many people swallow that story hook, line, and sinker.
Did you ever hear about the dac designer trick of adding a DC offset to the modulator so that modulator noise, which varies with DC level of the analog output signal, is minimized when when music is silent? There are other things too. Several years ago ESS tried to explain about of few of them. Unfortunately, people who have been indoctrinated by Amir somehow believe he knows more than ESS does about dacs. As a result most of those people just reject what ESS said, while claiming that ESS did not provide "proof." Look, whether ESS provided enough proof or not, that doesn't automatically make Amir right. He hasn't provided "proof" either.
Last edited:
If I tried my open back HD600 vs my closed back Momentum MkIIs with ANC, there's no difference in age for me.
However, I do notice a lot of tones in the sweep near the end of my spectrum that are much lower and moving around in the Momentum. Similar to but quite a lot worse compared to a sweep like this on good old cassette tape. This was on Bluetooth, so that might have something to do with it.
The ADC’s anti-aliasing filter cut-off frequency for CD is typically 20kHz. Which means that the resulting lowest DAC image-band frequency for CD is [44.1kHz - 20kHz = 24.1kHz], which is obviously ultrasonic. That’s the LOWEST image frequency. About all music spectral content occurs much lower than 20kHz, however, so DAC image frequencies of music (not simply audio tones) consequently manifest much higher.
For example, a 1kHz music component would produce a lowest DAC image frequency of [44.1kHz - 1kHz = 43.1kHz], which is far into the human ultrasonic. It’s quite unlikely that anyone could hear a 24.1kHz image frequency, let alone hear the image frequencies produced by the great bulk of musical program content, and which manifest ABOVE 24.1kHz.
Last edited:
Any evidence young girls are more likely to hear frequencies above 20kHz than young boys? I know females have better 2kHz to 8kHz hearing and makes better 1kHz to 2kHz but not seen anything to suggest girls are more likely to hear above 20kHz.
I agree, I deleted that notion out of my comment because while it’s something we’ve always heard tell, I’ve never seen published evidence of that. It seems to me that there is no physiological reason why the hearing acuity of a young boy and a young girl (meaning, pre-pubescent) should significantly differ, but I’ve zero expertise in that area.
Aren't those calculations a bit approximate, given that some ADC filters have a slow-ish roll off in the transition band?
Yes, they are. FIR anti-alias filter transition-band slopes do vary a bit, as does their stop-band suppression level. It makes for an easier to understand explanation, though, simply to use a typical anti-alias filter pass-band cut-off frequency of 20kHz. Even should we assume a cut-off frequency of 21kHz, however, that still locates the lowest image frequency at an ultrasonic, 23.1 kHz. Ultimately, the filter’s full suppression has to be reached by 22kHz, of course, or aliasing could result.
Last edited:
"Under ideal laboratory conditions, humans can hear sound as low as 12 Hz and as high as 28 kHz..."
https://en.wikipedia.org/wiki/Hearing_range
Also, I know I used to hear some ultrasonic burglar alarm transducers although I don't know what SPL they produce. You could walk around a room with those things on and hear the comb filtering reflections causing rippling amplitude variations in the sound according to movement around listening position in the room.
https://en.wikipedia.org/wiki/Hearing_range
Also, I know I used to hear some ultrasonic burglar alarm transducers although I don't know what SPL they produce. You could walk around a room with those things on and hear the comb filtering reflections causing rippling amplitude variations in the sound according to movement around listening position in the room.
Last edited:
But weren’t those detection thresholds at something like, 90-100dB louder than midrange frequency detection thresholds?
Which would seem to make sense, as the ear has a roll-off slope, like any other analog filter. If that slope is down 100dB @ 28kHz, boosting the signal by that same amount should make that frequency detectable, I would think.
Which would seem to make sense, as the ear has a roll-off slope, like any other analog filter. If that slope is down 100dB @ 28kHz, boosting the signal by that same amount should make that frequency detectable, I would think.
True. However, those curves like all hearing thresholds are based on averages with a limited number of test subjects. Moreover, back when that work was done it was much more common for researchers to discard "outlier" data.
However, overall I would tend to agree that in most cases if people hear effects of ultrasonic (and or RF) frequency content, its probably due to some resulting lower frequency content produced by the reproduction system. For one thing, it hasn't been that many years since manufacturers starting designing opamps that are more tolerant of RF (such as that produced by unfiltered dacs).
LM4562 (aka LME49720) being some of the last without protection. Even with more tolerant opamps, clock edge noise may go right around an opamp through an integrating cap.
Even with filtering, I have measured square wave RF closely related to the BCLK frequency at the output of a dac I/V stage. It was hard to measure, but a fancy scope with dot-averaging and triggering off the dac clock could image the square waveform buried in the modulator shaped-noise. I then connected an HPA with very high gain to the dac output and measured the same small-ish RF square wave at the headphone amp output jack (and, yes, I checked to make sure the RF at the HPA output wasn't due to stray coupling of radiated RF or some similar effect).
However, overall I would tend to agree that in most cases if people hear effects of ultrasonic (and or RF) frequency content, its probably due to some resulting lower frequency content produced by the reproduction system. For one thing, it hasn't been that many years since manufacturers starting designing opamps that are more tolerant of RF (such as that produced by unfiltered dacs).
LM4562 (aka LME49720) being some of the last without protection. Even with more tolerant opamps, clock edge noise may go right around an opamp through an integrating cap.
Even with filtering, I have measured square wave RF closely related to the BCLK frequency at the output of a dac I/V stage. It was hard to measure, but a fancy scope with dot-averaging and triggering off the dac clock could image the square waveform buried in the modulator shaped-noise. I then connected an HPA with very high gain to the dac output and measured the same small-ish RF square wave at the headphone amp output jack (and, yes, I checked to make sure the RF at the HPA output wasn't due to stray coupling of radiated RF or some similar effect).
Last edited:
Have you ever told a youngster he can't do something and then he does. Definitive proof that young boys are inherently deaf.
The same holds for the reconstruction/interpolation filter of a DAC and imaging, but as far as I know, almost all ADC and DAC chip manufacturers actually use half-band filters that are only 6.02 dB down at 22050 Hz and that reach full suppression at 24.1 kHz.
Then again, there is a good chance that the recordings on a CD were made at a higher sample rate and were then decimated to 44.1 kHz. The filter used for that decimation step is what matters then.
IIRC, SRC4392 ASRC chip uses a half-band filter.
That said, modern dac and ADC chips from AKM and ESS usually have selectable filters, some of which roll-off pretty fast in the transition band. Don't know what ESS uses for its internal ASRC though.
That said, modern dac and ADC chips from AKM and ESS usually have selectable filters, some of which roll-off pretty fast in the transition band. Don't know what ESS uses for its internal ASRC though.
It does, that's why in my valve DAC, I let the FPGA interpolate by two before sending the signal to the SRC4392.
Isn't the last decimation step for a CD recording normally done in software?
So if all we're down to here is filter effects, doesnt it rather point to measurements being what matters? And changing caps, op-amps etc isnt going to be audible as there is no measurable difference when these things are done?
Which specific measurements? Modulator noise as a function of DC offset? Other audio signal correlated noise? Random noise jitter?
Incorrect. Amir hasn't proven he measures everything that matters.
--------------------
In another thread you wrote:
We can tell you are not technical. Yet you really, really want to believe in Amir's unproven theory. Why? Is it that you want to believe highly technical stuff is really very simple? Like, no education needed? If so, you would be mistaken.
Last edited:
- Status
- Not open for further replies.