As I promised in this thread, I have prepared a little test to try to determine if the phasing of harmonics plays a role in the perception of Xover distortion.
Test files are in the zipped folder attached to this post.
The test consists of a 1KHz sinewave affected by Xover dist. The amount of distortion is severe, but not caricatural: the transistors are just underbiased.
The resulting sound is highly strident and unpleasant; no surprise there.
I have then created a second signal, completely similar, except the phases of the harmonics are reshuffled completely randomly (based on a random number generator).
Can you spot a difference, and if yes, can you tell the "natural" sound from the "artificial" one?
Please, try to refrain from cheating, and don't look at the .wav files, just listen to them.
Personally, at first I heard no difference (on my lousy PC speakers).
I then tried with headphones, and there I could reliably detect a more strident version.
I will probably repeat the test at other frequencies
Test files are in the zipped folder attached to this post.
The test consists of a 1KHz sinewave affected by Xover dist. The amount of distortion is severe, but not caricatural: the transistors are just underbiased.
The resulting sound is highly strident and unpleasant; no surprise there.
I have then created a second signal, completely similar, except the phases of the harmonics are reshuffled completely randomly (based on a random number generator).
Can you spot a difference, and if yes, can you tell the "natural" sound from the "artificial" one?
Please, try to refrain from cheating, and don't look at the .wav files, just listen to them.
Personally, at first I heard no difference (on my lousy PC speakers).
I then tried with headphones, and there I could reliably detect a more strident version.
I will probably repeat the test at other frequencies
Attachments
Last edited:
First time I've seen this but I'll give them a listen.
(Elvee, these file hosting sites are a nightmare, maybe that's why you have had no takers... this ones average... slow, unintuitive, asks for codes, asks you to create links that are only valid for so long, have to wait before you can down load the next file 😀 hint maybe use Dropbox next time 🙂)
(Elvee, these file hosting sites are a nightmare, maybe that's why you have had no takers... this ones average... slow, unintuitive, asks for codes, asks you to create links that are only valid for so long, have to wait before you can down load the next file 😀 hint maybe use Dropbox next time 🙂)
Elvee, I can differentiate these easily on my laptop speakers every single time and never get them mixed. They are absolutely totally different to listen to. The first is smoother and less harsh by many miles.
Yeah, they do sound different on my laptop speakers.
Are we testing phasing of distortion components, or hard clipping ???.
Dan.
Are we testing phasing of distortion components, or hard clipping ???.
Dan.
I've only just peeked at the files in Audacity now and can see they are both severely clipped... I'm not so sure they are meant to be tbh.
Elvee 🙂 what are we listening to here ?
Elvee 🙂 what are we listening to here ?
Well I made a few mistakes for sure, but I learnt much from Pavel on producing files, and one of the main things being that nothing gets past you guys and your file analysis.
This is a classic example of a nonlinearity converting changes in harmonic phase into changes in amplitude. The hard clipping makes the phase audible.
I've been bitten once by dropbox, which is why I now use obscure, inscription-free services; the downside is, they are rather clumsy and impractical....
Well, sorry guys, but something did go wrong
LTspice played tricks on me and both files ended up clipped, which was not supposed to be part of the test.
I'll have to recreate the files from scratch, since LTspice crashed in an unrecoverable way. First time it happens.
In the mean time, if someone has a better suggestion for a hosting site, it is welcome (not dropbox, and inscription-free)
Notice that the harmonic amplitudes are now different.
I'm labouring this point because I think many (all?) of the claims of phase audibility out there are actually based on mistakes of this kind.
LOL.....it can happen to the best of us 😉.
The 6 P's......Prior Preparation Prevents a P*** Poor Performance... 😎.
Dan.
I'm not sure that I agree.
Change in timing of harmonic components subtly changes the timbre.
Dan.
Don't jump to conclusions based on these files: they were corrupted in a way that negated the prime purpose of this test, ie. having exactly the same spectral content regarding magnitude.
I am currently generating a (hopefully) correct set of files.
Last edited:
Ok, good.
I have contemplated exactly this kind of testing...I just haven't extracted the digit in order to do so....kudos from me to you for trying
.Dan.
OK, here are the new links (through a dodgy storage provider again, beware of the traps, hint: the kapcha should be "woop"):
Free - Envoyez vos documents
Free - Envoyez vos documents
Note that I have kept the same file names to remain consistent with the poll options, but this doesn't mean they correspond to the same content (you can cheat anyway, it is thus immaterial, but try to remain honest, there is valuable data to be gathered if you are)
Free - Envoyez vos documents
Free - Envoyez vos documents
Note that I have kept the same file names to remain consistent with the poll options, but this doesn't mean they correspond to the same content (you can cheat anyway, it is thus immaterial, but try to remain honest, there is valuable data to be gathered if you are)
Google Translate does not work with those links.
Perhaps try a different service as per suggestions.....the previous one required registration.....I'm losing interest really fast 😡.
Dan.
Perhaps try a different service as per suggestions.....the previous one required registration.....I'm losing interest really fast 😡.
Dan.
That is exactly the point I'm trying to make. Nonlinearity converts changes in harmonic phase into changes in spectral content. I believe that most tests for phase audibility are broken because of nonlinearity in the playback systems used. This one is just an extreme example.
Ultimately if you crank the playback level high enough, nonlinearity in the ear itself will do it.
- Status
- Not open for further replies.