0.01% i can hear at the highest frequency of the sweep that for sure i can say i can hear. 0.001% i belive i'm starting to have placebo effect, imaginary very very low beeps. Earphones, a bit late to play on speakers atm.
1% and 0.1% were clearly audible on my old HP Probook 840 with some cheap Pioneer headphones, though 0.1% started to become clearly audible only closer to end of the sweep. 0.01% I couldn't detect at the levels of 1% and 0.1% but with increased volume it definitely got audible but at the same time the headphones started to distort a whole lot more too, with strong 2nd and 3rd harmonic in the mix especially at the lower frequencies.
Laptop's output is actually very dirty too, with significant "digital" noises in background, not just hiss so it isn't a particularly good thing to test with but for the time being it is all I have. Later in next week I'll be back home from my trip to another country and can do proper tests on my main sound system out of a clean source. I'm pretty sure 0.001% will not be audible, except maybe at much elevated volumes.
Neat test though, but there's some room for improvement as the switching is a bit sudden and causes clicks which can induce the placebo effect aspects. I could clearly hear those clicks in the 0.001% test but not the actual extra harmonic between those clicks.
Laptop's output is actually very dirty too, with significant "digital" noises in background, not just hiss so it isn't a particularly good thing to test with but for the time being it is all I have. Later in next week I'll be back home from my trip to another country and can do proper tests on my main sound system out of a clean source. I'm pretty sure 0.001% will not be audible, except maybe at much elevated volumes.
Neat test though, but there's some room for improvement as the switching is a bit sudden and causes clicks which can induce the placebo effect aspects. I could clearly hear those clicks in the 0.001% test but not the actual extra harmonic between those clicks.
Interesting. The clicks are no higher in amplitude than the harmonic, but cover a broad spectrum. I would say that you can hear 0.001%.Tiido said:
Ed
At 0.1% I can hear it clearly, at 0.01% the higher pitched beeps have disappeared. Im using Hifiman Sundara straight from lenovo thinkpad atm. Im curious to try different harmonics or the IMD test. My latest pass-inspired class A amp does 0.05%THD close to full power. I feel like I can hear the distortion at >100dB listening levels though thats probable more IMD.
Last edited:
I did not post the IMD test because it is potentially hazardous to tweeters. 😉Rupopulles said:
Ed
Yeah, it has to do with the broader spectrum, it can excite more of those hairs in the cochlea. I counted this as not valid for this test and voted 0.01%
I finally did the test with my proper sound setup and this time 0.01% didn't need any elevated volume to discern although it was very very faint. I also heard the tone over much larger part of the sweep compared to the laptop+headphones (on 0.1% too), I was using my OPPO BDP105, which has a proper DAC and definitely good enough analog stages, speakers are two pairs of soviet era boxes Amfiton 50A022 and 100AS022 and amplifier my homemade thing built over the years.
0.001% was not heard at all though, it is much below ambient levels in the room and buried deep into the noisefloor of the amplifier. I even went right to the speakers to counteract noise from outside (sea, wind, birds lol) but all it achieved was making the slight hiss from tweeters get really clear lol.
I will try 0.001% again when a friend comes to visit next month with his fancy headphones. I will do a test run with my homemade headphone amplifier which has pretty much inaudible noisefloor and some premium spec silicon used in its signal path, perhaps it can make a difference.
I have held 0.03% figure as "good enough", but perhaps I should go a bit lower than that now lol. This has been quite enlightening, thänk you for making this ~
0.001% was not heard at all though, it is much below ambient levels in the room and buried deep into the noisefloor of the amplifier. I even went right to the speakers to counteract noise from outside (sea, wind, birds lol) but all it achieved was making the slight hiss from tweeters get really clear lol.
I will try 0.001% again when a friend comes to visit next month with his fancy headphones. I will do a test run with my homemade headphone amplifier which has pretty much inaudible noisefloor and some premium spec silicon used in its signal path, perhaps it can make a difference.
I have held 0.03% figure as "good enough", but perhaps I should go a bit lower than that now lol. This has been quite enlightening, thänk you for making this ~
IME HD/THD is most audible when listening for IMD in a familiar sound. As humans we tend to notice small nuances of expression in vocals. Vocal harmonies produce beating, a type of envelope modulation dependent on the musical interval or intervals summing together. The texture of that composite sound can be very revealing of low level of distortion.
OTOH, IME hearing HD in a steady state sine wave is a different task for the human ear/brain system. For steady state test tone results to be extensible to real music would require that the human auditory system is LTI (linear, and time-invariant) and that it is stationary (in the statistical sense). If you look at something like an EEG of brain activity when listening to speech or music, the auditory system appears to be non-LTI and nonstationary. Non-LTI and nonstationary is also true for EKG signals, which further tends to illustrate that biological systems may be far more complex than human engineered systems.
OTOH, IME hearing HD in a steady state sine wave is a different task for the human ear/brain system. For steady state test tone results to be extensible to real music would require that the human auditory system is LTI (linear, and time-invariant) and that it is stationary (in the statistical sense). If you look at something like an EEG of brain activity when listening to speech or music, the auditory system appears to be non-LTI and nonstationary. Non-LTI and nonstationary is also true for EKG signals, which further tends to illustrate that biological systems may be far more complex than human engineered systems.
Tiido - Thanks! 0.001% is below the noise floor of 16-bit audio. The sweeps have noise added to preclude quantization error.
Ed
Ed
Interesting.PC speakers and an Xfi. I did select the speakers carefully after a fashion when monitor had to go. So 1% easily but noticed an effect that declined as distortion reduced, 😉 It reminded me of a party trick that annoyed some when young. The BBC used to stop broadcasting at 11pm (or some similar time) and just broadcast a tone to get people to turn the set off. For some reason I copied it vocally and found that if I shifted the tone a touch it made it far more irritating for me and everyone else. I suspect is caused resonances in the inner ear. Reason a bit like rf mixing. 2 frequencies produce sum and difference frequencies. Can't think of any other explanation. 😉 Maybe you can. I noticed some strange oral tone effects on all of the sections generally at the higher frequency end.