How much distortion can you hear? Take this two-minute test to find out.
The attached file contains four frequency sweeps. Each sweep consists of a fundamental from 100-500Hz and a seventh harmonic at 1%, 0.1%, 0.01%, and 0.001% respectively.
The test should be made with good speakers or headphones (computer speakers may not be able to reproduce the fundamental).
I made the harmonic easy to detect by switching it on and off at a 1Hz rate and putting it in the frequency range where the ear is sensitive.
Ed
The attached file contains four frequency sweeps. Each sweep consists of a fundamental from 100-500Hz and a seventh harmonic at 1%, 0.1%, 0.01%, and 0.001% respectively.
The test should be made with good speakers or headphones (computer speakers may not be able to reproduce the fundamental).
I made the harmonic easy to detect by switching it on and off at a 1Hz rate and putting it in the frequency range where the ear is sensitive.
Ed
Kudos to the people who voted. 🙂
The room has to be quiet in order to hear the harmonic at 0.1%. This may actually be a test of the room.
I debated whether I should vote for 0.01%. At that level, I cannot hear a tone but rather perceive that something has been added.
The last sweep at 0.001% sounds clean to me. The harmonic is below the noise floor of 16-bit audio.
All sweeps have noise added to avoid quantization error.
Ed
The room has to be quiet in order to hear the harmonic at 0.1%. This may actually be a test of the room.
I debated whether I should vote for 0.01%. At that level, I cannot hear a tone but rather perceive that something has been added.
The last sweep at 0.001% sounds clean to me. The harmonic is below the noise floor of 16-bit audio.
All sweeps have noise added to avoid quantization error.
Ed
I am impressed by the people who can hear 0.01% distortion. That requires a very quiet room and a louder-than-usual listening level.
Ed
Ed
In general, the question should not be asked in this way - check out post #2 under
https://www.diyaudio.com/community/...st-possible-thd-n-really-the-best-way.367692/
Furthermore, it can be taken for granted that there is no amplifier unit in real live, that only the 7th harmonic is contained in a THD+N value.
I designed the THD test to make low levels of distortion easily detectable. The test also has to work on all systems. I was not trying to model a real distortion spectrum.
I learned that some people can hear 0.01% THD.
BTW, I also have an IMD test. It is unpleasant to listen to.
Ed
I learned that some people can hear 0.01% THD.
BTW, I also have an IMD test. It is unpleasant to listen to.
Ed
Ha. A curse and a blessing being Autistic. 😉 I prefer seeing it as a blessing and use it to help others (ex. at church we had a feedback during practice only I could hear, though I am sure my Autistic friends also could hear it, that if not heard still affect sound quality).
Now would I only buy equipment at this distortion level? No, because I wouldn't be listening for distortion. 0,01% be good enough. 😉
Something strange going on here
With my phone speaker, the first two were easy to discern, but it was much harder with my headphones. Are we just hearing some kind of IMD on the playback device?
Brian
With my phone speaker, the first two were easy to discern, but it was much harder with my headphones. Are we just hearing some kind of IMD on the playback device?
Brian
I can hear 0,01% in my Beyer DT770 headphones.. but just barely, and only at the highest frequencies of the sweep at the end there.
Brian - The phone speaker is unable to reproduce the fundamental. The test is invalid on the phone.
StigErik - I expect that headphones generally do better than loudspeakers.
Ed
StigErik - I expect that headphones generally do better than loudspeakers.
Ed
I will try it on speakers as well ! My Magnepans have super low distortion, so it’s gonna be interesting …
I accept that for the lower frequencies, but it is clearly audible all the way up. I don’t think it’s a big deal as most people won’t do the test that way, but it’s interesting nonetheless
Brian
Listening to distorted sine waves has little to do with distorted music perception. Extrapolating from sine waves to more complex waveforms is a model based on the assumption that the human auditory system LTI (linear time invariant) and stationary for FR and for HD. We might add to the model: insensitive to phase, except sensitive to phase at LF. The model kind of works for some time-domain waveforms and not so well for others. In particular, IME it is sometimes easier to hear IMD in well recorded vocal harmonies, where a cordal 'texture' may be affected by the underlying nonlinearity. Please note that the term 'texture' in relation to a sound is sometimes used in the field of 'Auditory Scene Analysis,' which in turn is the study of auditory perception in the presence of noise (the term 'noise' including sounds such as those found in a noisy restaurant).
Agreed.
My goal was to make low levels of distortion easily detectable so that equipment designers can set worst-case targets.
Ed
Try out this link ... think this is a bit more realistic 😉
http://abx.digitalfeed.net/list.lame.html
I got the LAME 128 kbs 88% right, but at 256 kbs I'm down to around 50% ....
(used Bayer Dynamcis DT770 Pro on a E-MU Tracker pro)
(And have had my hearing measured a good number of times through my work + plus listening test of various new DSP processing algorithms)
I know it is not a pure THD test, but it made me wonder how far we need to go on achieving super low distortion figures .... don´t really think it matters so much below say 0.1% as so many other factors, not least the room
The good thing about this sort of ABX test is it takes the guessing out ....
Give it a try 🙂
... and sorry for going off track or hijacking the thread
http://abx.digitalfeed.net/list.lame.html
I got the LAME 128 kbs 88% right, but at 256 kbs I'm down to around 50% ....
(used Bayer Dynamcis DT770 Pro on a E-MU Tracker pro)
(And have had my hearing measured a good number of times through my work + plus listening test of various new DSP processing algorithms)
I know it is not a pure THD test, but it made me wonder how far we need to go on achieving super low distortion figures .... don´t really think it matters so much below say 0.1% as so many other factors, not least the room
The good thing about this sort of ABX test is it takes the guessing out ....
Give it a try 🙂
... and sorry for going off track or hijacking the thread
I can just about hear 0.01% at higher volumes.I tried on 4 18 inch subs and 2 15 inch 3 ways.I can only hear it when the fundamental dips down due to room modes,so room acoustics probably matter a lot more.I would be curious about lower order harmonics,but those would be probably even harder to notice,i'm going to boot up audacity and try tones a octave apart at 10 % and 1%,but i'm not sure i'll hear it to be honest