notch out the 11khz peak until it sounds fine, sweep around to confirm its that or sometging else.
This kind of EQ trick probably changes / needs tweaking track by track, which might seem pointless. Its not pointless in my opinion though, because by experimenting with the EQ you learn listening. You might find out some setting that works for you better than some average target curve 🙂 or you might find out some music indeed has some weird sound to them, for some odd reason. You might learn that EQ is something better left alone, perhaps find the track in lossless format, perhaps that works best. What ever it is you are still learning and getting experience.
You might disable the eq after a while, test something else on some other day and track. Its all fine, more hours on listening and looking at graphs and all that should develop listening and other skills and lead to better and better sounding system eventually 🙂 Important thing is that you are now zoned in to this particular phenomenon so might as well use the opportunity to play around with it, get familiar with the phenomenon you perceive.
Comments from krivium or fluid or wesayso or anyone else who likely have lots of listening and tweaking various systems are golden, but it is very hard to connect comments of others to actual sound you perceive. Listening and tweaking yourself is the way to go, and develop skills and opinion, build up experience. Have fun! 🙂
You might disable the eq after a while, test something else on some other day and track. Its all fine, more hours on listening and looking at graphs and all that should develop listening and other skills and lead to better and better sounding system eventually 🙂 Important thing is that you are now zoned in to this particular phenomenon so might as well use the opportunity to play around with it, get familiar with the phenomenon you perceive.
Comments from krivium or fluid or wesayso or anyone else who likely have lots of listening and tweaking various systems are golden, but it is very hard to connect comments of others to actual sound you perceive. Listening and tweaking yourself is the way to go, and develop skills and opinion, build up experience. Have fun! 🙂
Last edited:
Eq at 11khz to increase 'air'.
Look at the 5khz and 6,5khz peaks, this is usually where it is.
Yes better speakers makes artefacts more obvious. But i agree with Tmuikku it shouldn't stop you to be into the music.
Tmuikku, please don't associate me with anything golden or Fluid or Wesayso ( those guy are waaayyy more advanced i'll ever be about what we talk in here).
My advice or pov have no more weight than anyone else, we all learn from each others and share our experience.
Mind you i consider your pov as golden ones on a number of things... 😉
More seriously it makes me feel strange.
Look at the 5khz and 6,5khz peaks, this is usually where it is.
Yes better speakers makes artefacts more obvious. But i agree with Tmuikku it shouldn't stop you to be into the music.
Tmuikku, please don't associate me with anything golden or Fluid or Wesayso ( those guy are waaayyy more advanced i'll ever be about what we talk in here).
My advice or pov have no more weight than anyone else, we all learn from each others and share our experience.
Mind you i consider your pov as golden ones on a number of things... 😉
More seriously it makes me feel strange.
It is good to check the content of a song that bothers you. Then you know if it is in the source material.
I would check from one of the measurements of your waveguide the waterfall plot for a resonance. I am using a throat adapter with a continuous tubing and it causes a resonance at the frequency that matches the diameter. The resonance leaves a trace in the frequency response, but nothing that audibly bothers me.
After this, I would experiment with room EQ curves and get to know your subjective impression. You need to familiarize yourself with what you hear when you make changes, otherwise more complex technical approaches might fail.
I have implemented a very gentle sloping high shelf after @hifijim's suggestion, f=1k, Q=0,34, Gain=-1.5 dB, which is much less than the suggested negative gain. I also have some low Q PEQs that raise the response below 1k and effectively move the saddle of the slope to around 1k.
With this EQ, I hear the HF sibilance that you describe in the source material, but it does not bother me.
I would check from one of the measurements of your waveguide the waterfall plot for a resonance. I am using a throat adapter with a continuous tubing and it causes a resonance at the frequency that matches the diameter. The resonance leaves a trace in the frequency response, but nothing that audibly bothers me.
After this, I would experiment with room EQ curves and get to know your subjective impression. You need to familiarize yourself with what you hear when you make changes, otherwise more complex technical approaches might fail.
I have implemented a very gentle sloping high shelf after @hifijim's suggestion, f=1k, Q=0,34, Gain=-1.5 dB, which is much less than the suggested negative gain. I also have some low Q PEQs that raise the response below 1k and effectively move the saddle of the slope to around 1k.
With this EQ, I hear the HF sibilance that you describe in the source material, but it does not bother me.
You've got some track record in listening while tweaking knobs if I've understood 🙂
I like to think, while mixing a song, that one needs to shave the mix out gradually. Balance the tracks, tweak here, tweak there, remove the previous tweaks, rebalance, and soon its there. This back and forth tuning somehow helps brain to hear how things relate to each other. In the end there aren't necessarily that many tweaks left on the final mix, but all the steps were necessary to be able to arrive to the end result. The tweaking allows hearing to find out whats what, zone in.
Similar thing with loudspeaker systems, do this, do that, hustle around with interesting stuff and suddenly you start to get handle on what is what, spot new things, concentrate on those and get familiar with. Soon one can relate perceived sound to graphs and vice versa. At least to some extent. The listening and tweaking procedure is important.
If there was just a sound without ability to tweak it the brain gets used to it and you would be happy. Then one day, for some reason, you woukd notice something is wrong and start diy speaker hobby hoping to do something about it. There might be multiple things sub optimal but you can't pinpoint what they are or how to better. As soon as one can alter the sound/setup somehow, preferably in multiple ways, start tweaking things it allows to get familiar with the sound from multiple angles and get to know it better. Suddenly you'll spot a problem and can fix it, then another, then another ad so on, gradually untangling it 🙂
Last edited:
@HeadShake demonstrated to me that "sss" and "shhh" sibilance could be greatly affected by the baffle size/shape immediately around the tweeter. A very small baffle was much better than a larger baffle, and bare no-baffle tweeter was best... The responses were eq'd to be the same on-axis. The biggest measured difference in three tweeter baffles was in the CSD plots. The smaller the baffle, the more clean the decay in the first 1 ms. This experiment convinced me that sibilance error is not simply a frequency response phenomenon.
j.
j.
Tmuikku,
Yes of course i'm not a newb with the tools and my ears but i find designing/tweaking loudspeakers to be different than working with audio.
There is hidden things, or things which i have not learned to identify by ear... like what Jim is talking about: which happen in time domain (like stored energy).
I'm more and more convinced that time domain is maybe more important than frequency domain for realism.
Hifijim, you talk about dome tweeter?
Yes of course i'm not a newb with the tools and my ears but i find designing/tweaking loudspeakers to be different than working with audio.
There is hidden things, or things which i have not learned to identify by ear... like what Jim is talking about: which happen in time domain (like stored energy).
I'm more and more convinced that time domain is maybe more important than frequency domain for realism.
Hifijim, you talk about dome tweeter?
Yes. @HeadShake invited me to his place to do some listening evaluations of various dome tweeters he had purchased. One of the best sounding tweeter options we listened to was an inexpensive Audax tweeter with a very small flange, something like 35 mm. He had it suspended in free space, on the end of a rod. It sounded very very good. Very three dimensional, very detailed. It changed my perspective a bit.
https://www.diyaudio.com/community/threads/headshakes-far-field-3way.382393/page-7#post-7006250
https://www.diyaudio.com/community/threads/headshakes-far-field-3way.382393/page-7#post-7006250
Can you run plugins?
https://plugins4free.com/plugin/537/
It's a 32 bit VST plugin De-Esser, I've used this long ago making off-line tweaks to tracks that needed it for use in my car. (ran in in a batch program on certain material)
It worked wonderful for some solo tweaks on tracks/albums that really needed some work. The fun part is that you can solo the "Sss" part itself and find the exact frequency that way.
My mid-side EQ story is a long one, and it gets.... complicated... Best start with the thread @krivium linked.
It completely solved my hunger for using tools like de-essers though 😉.
https://plugins4free.com/plugin/537/
It's a 32 bit VST plugin De-Esser, I've used this long ago making off-line tweaks to tracks that needed it for use in my car. (ran in in a batch program on certain material)
It worked wonderful for some solo tweaks on tracks/albums that really needed some work. The fun part is that you can solo the "Sss" part itself and find the exact frequency that way.
My mid-side EQ story is a long one, and it gets.... complicated... Best start with the thread @krivium linked.
It completely solved my hunger for using tools like de-essers though 😉.
Thank you Jim.
Ok i think there is a range of box width which can be an issue with not waveguided tweeters.
You have either no distance to edge or a 'big' distance to edge the possible induced time issues are either too short to be audible ( merged with direct signal) either long enough for the brain to integrate as distinct from direct signal (time delayed and lower in level).
Now it ask about what is 'big' distance? Here i think it depend of the xover frequency hence the relationship between wavelength at xover and box width...
I had multiple occasion to listen for longtime to inwall loudspeakers with not waveguided tweeters and never find them to be more bothersome related to sibilance.
For freestanding loudspeakers of course if you can include large roundover it's a plus ( or a slanted shape ).
I can testify my 3 ways (54cm width, xover initially at 4,5khz, raised to 6khz, slanted shape) doesn't exhibit more sibilance than my other dome based tweeter nearfield ( Alesis monitor one passive and monitor1mk2 active).
With waveguided tweeters it's a non issue in my experience ( Genelec 1037, Atc Scm110A). Same with horn loaded monitors i met.
Ok i think there is a range of box width which can be an issue with not waveguided tweeters.
You have either no distance to edge or a 'big' distance to edge the possible induced time issues are either too short to be audible ( merged with direct signal) either long enough for the brain to integrate as distinct from direct signal (time delayed and lower in level).
Now it ask about what is 'big' distance? Here i think it depend of the xover frequency hence the relationship between wavelength at xover and box width...
I had multiple occasion to listen for longtime to inwall loudspeakers with not waveguided tweeters and never find them to be more bothersome related to sibilance.
For freestanding loudspeakers of course if you can include large roundover it's a plus ( or a slanted shape ).
I can testify my 3 ways (54cm width, xover initially at 4,5khz, raised to 6khz, slanted shape) doesn't exhibit more sibilance than my other dome based tweeter nearfield ( Alesis monitor one passive and monitor1mk2 active).
With waveguided tweeters it's a non issue in my experience ( Genelec 1037, Atc Scm110A). Same with horn loaded monitors i met.
Yes, but how successfully is always a question. It has it's own noise reduction routine or it can use VST Plugins to process the audio.
Your little finger is correct, although I'm not sure how well that translates from French to English 😉
The basic description is a technique to try and address interaural cross talk, head shadowing, HRTF are terms that may be familiar.
As wesayso said before with speakers or rooms that have reduced early reflections this becomes much more apparent than in those systems that have higher levels, as the reflections fill in the crosstalk making it harder to hear. This might seem like early reflections are good but they are not without their own problems.
A stereo signal can be converted to mid and side components through a sum and difference process, all the content that is the same in both channels like centre panned becomes the mid signal and things that are different becomes the side signal. These signals can then be processed separately before being recombined into a left right output. Applying an EQ that is like a correction for the crosstalk to the mid signal and rebalancing the relative levels of the mid vs side has a useful effect. The tonality shift between dead centre and off centre can be corrected. Like most things it is not without a down side but it is possible to find a useful compromise.
This is another fairly deep rabbit hole to go down, an interesting one though.
I would be cautious about the distinction between EQ ing the speaker and room and EQ ing for specific tracks. Remastering for specific tracks can be effective but that same EQ applied to all tracks is not going to work. This is one of the circle of confusion difficulties when trying to assess speakers using already recorded music. It becomes important to listen to a wide range of source material to be sure the EQ wasn't just compensating for a specific tracks issues. I have fallen victim to this many times myself. "That's it, I've got it, this sounds awesome", only to change tracks and have it sound awful.
@wesayso: I will try the VST.. 🙂
@fluid: The mid/side EQ concept seems a little hard to understand for me at the moment... I will try out the simpler things like De esser and see if it is of any help. 🙂
In the meanwhile, I tried to plot the burst decay of the compression driver on the horn (4ms gated measurement). Here is how it looked like:
Does it tell anything about a resoance in the 3kHz region.. 🤔
@fluid: The mid/side EQ concept seems a little hard to understand for me at the moment... I will try out the simpler things like De esser and see if it is of any help. 🙂
In the meanwhile, I tried to plot the burst decay of the compression driver on the horn (4ms gated measurement). Here is how it looked like:
Does it tell anything about a resoance in the 3kHz region.. 🤔
Try the Burst Decay within the spectrogram view I think that will be easier to interpret.
Hi,
M/S processing and treatments are powerful tools if you know how to use them and why to use them.
It's a matricing ( from L/R signal you extract mid and side information).
To better understand it Vineeth* make a search about M/S microphone couple technique it should help you to get the principle at work.
From there we could start discussing what can be done with it.
If you think de-esser is easy to apply... you'll have fun! 😉
Yes there is something happening in the 2/3khz range.
A resonance, a stored energy thing which release it with a bit of delay relative to the initial message.
There is something happening at ~5khz and abit under 6khz as well as~7,2khz and i would look at them about sibillance ( but it's not this severe from the graph imo, that said audibility can be different, our brain can be difficult to trick).
Fluid if the translation of 'mon petit doigt m'a dit' is as absurd and a non sense in english as it is in french then, it's all good and you have the meaning! Lol.
A nice way to not give an answer about something which wasn't asked... 🙂
*Vineeth it occured to me i use this part of your avatar as it is your firstname and a shortcut, i hope it is the case and i'm not rude to you. If that is not the case please tell me and i will use your complete avatar, and apologize. 😉
edit: it seems i switched italic on and won't go back to regular... 'quel boulet!'
M/S processing and treatments are powerful tools if you know how to use them and why to use them.
It's a matricing ( from L/R signal you extract mid and side information).
To better understand it Vineeth* make a search about M/S microphone couple technique it should help you to get the principle at work.
From there we could start discussing what can be done with it.
If you think de-esser is easy to apply... you'll have fun! 😉
Yes there is something happening in the 2/3khz range.
A resonance, a stored energy thing which release it with a bit of delay relative to the initial message.
There is something happening at ~5khz and abit under 6khz as well as~7,2khz and i would look at them about sibillance ( but it's not this severe from the graph imo, that said audibility can be different, our brain can be difficult to trick).
Fluid if the translation of 'mon petit doigt m'a dit' is as absurd and a non sense in english as it is in french then, it's all good and you have the meaning! Lol.
A nice way to not give an answer about something which wasn't asked... 🙂
*Vineeth it occured to me i use this part of your avatar as it is your firstname and a shortcut, i hope it is the case and i'm not rude to you. If that is not the case please tell me and i will use your complete avatar, and apologize. 😉
edit: it seems i switched italic on and won't go back to regular... 'quel boulet!'
Last edited:
I don't find the track you linked to be this bothersome regarding sss.
Let's go back to 1995 when my girlfriend of the time included a paradigm shift in my cultural habit, managing to drag me from concert halls full of beer loaded bearded furious people to Spanish's huge clubs in Cadaquès area full of non tattooed pretty girls wearing small skirts and dancing on cube... a bit of fresh air from the mosh pit i was used to. 🙂
At that time a remix of EBTG 'missing' was huge... and it was not pretty regarding sibillance! Mainly because of compression...
If you compare to the original track ( where the mixer used the same trick as in Massive Attack's 'Protection' but more discreet) it should be obvious there is difference.
I don't think it's an issue with the Todd Terry's remix, but a way to make the voice cut through the high hats and overall eq needed for the track to 'hit' in the low end as well as compression.
Another example from circa 2000 ( point in time when loudness war started to be embarrasing about quality) where you can hear an overprocessed female voice eqed and deessed ( because Nelly doesn't really sing -except on chorus- but whisper) to cut thru Timbaland heavy beats, i can clearly hear deessing taking place at time:
Let's go back to 1995 when my girlfriend of the time included a paradigm shift in my cultural habit, managing to drag me from concert halls full of beer loaded bearded furious people to Spanish's huge clubs in Cadaquès area full of non tattooed pretty girls wearing small skirts and dancing on cube... a bit of fresh air from the mosh pit i was used to. 🙂
At that time a remix of EBTG 'missing' was huge... and it was not pretty regarding sibillance! Mainly because of compression...
If you compare to the original track ( where the mixer used the same trick as in Massive Attack's 'Protection' but more discreet) it should be obvious there is difference.
I don't think it's an issue with the Todd Terry's remix, but a way to make the voice cut through the high hats and overall eq needed for the track to 'hit' in the low end as well as compression.
Another example from circa 2000 ( point in time when loudness war started to be embarrasing about quality) where you can hear an overprocessed female voice eqed and deessed ( because Nelly doesn't really sing -except on chorus- but whisper) to cut thru Timbaland heavy beats, i can clearly hear deessing taking place at time:
Last edited:
Thanks @krivium
I will look up information regarding this.. 🙂
I actually installed a deesser. The VST that wesayso linked was not working with Equalizer Apo. So I had to download another one which was supported in EQApo, but I couldn't find it making much difference in sibilance or anything else.. 😀 I have to see if it is some installation issue. However, a simple PEQ like the below was making much more noticeable differences in terms of 'hearable' differences in the song. Not much sibilance reduction, though.
I want to find out what is causing this. Maybe I will remove the horn and measure the bare driver and see if it is being caused by the horn attachment or something. Because I am not able to see the same energy release in the BMS4550 driver measurement (probably on some other horn) here:
https://www.justdiyit.com/grand-comparatif-de-compressions-1-pouce/4/
Actually, after listening to one of the songs that you linked above, my 'ssss' problem is not a problem at all in the sense that it occurs relatively less number of times in the song... 😀
It is just that everything sounds nice, and once in a while, when the 'ss' sound comes, my brain just gets hung up on that and loses focus on everything else going on in the song.
Please don't apologize. No need to use my avatar name.. My name is just 'Vineeth'. The rest all are just add-ons that came from my email id.. 😀
Everybody I know calls me as Vineeth only 🙂 (by the way, in Sanskrit, Vineeth means: Sweet, humble, polite etc 😀 )
Ok Vineeth,
The 'listen' button is here to help you identify which part you are going to treat: it gives you access to the 'sidechain' part of the compressor.
Deesser are selective compressor - not multiband process as was said previously: the compressor is triggered by sidechain signal ( a copy of input signal) on which either hp/lp filters are located to 'isolate' the offensive band. That way it does it's thing only on the ss. Sometimes there is eq involved too to exagerate the drive.
So by clicking on the 'listen button' and tweaking the lower parts parameters ( low and high are hp and lp, sensitivity the threshold above which the signal is amplified to drive the compressor, smooth...idk). Once setup you can play with reduction and attack/release to 'shape' the action.
The 'listen' button is here to help you identify which part you are going to treat: it gives you access to the 'sidechain' part of the compressor.
Deesser are selective compressor - not multiband process as was said previously: the compressor is triggered by sidechain signal ( a copy of input signal) on which either hp/lp filters are located to 'isolate' the offensive band. That way it does it's thing only on the ss. Sometimes there is eq involved too to exagerate the drive.
So by clicking on the 'listen button' and tweaking the lower parts parameters ( low and high are hp and lp, sensitivity the threshold above which the signal is amplified to drive the compressor, smooth...idk). Once setup you can play with reduction and attack/release to 'shape' the action.
Because this includes the waveguide I would equalize it flat or close to flat and look again. The peak of the waveguide gain can distort the relative levels.