So, at the moment, I do not have a measurement electret microphone. Unfortunate, as I really want one, but will have to wait a while, even though they're pretty cheap, as I am flat broke.
I am running ubuntu studio, and have the pulseaudio multiband eq always on.
Now, to even out the sound of my "orcus" fullrange speakers, I have been tweaking the EQ to get it to sound subjectively good for the music I listen to. But I wanted something a little less subjective.
So, I made a "test" file using Audacity to create a bunch of tones, each one at the respective band that the pulseaudio multiband eq uses.
Then, I tweaked the EQ so that each band had the same subjective loudness, to my ears.
Is this an acceptable method? Or do I need to account for the logarithmic perception of human hearing? If so, is there a simple way of doing this? I'm fairly math-savvy, calculus is not a problem, etc.
Thanks!
I am running ubuntu studio, and have the pulseaudio multiband eq always on.
Now, to even out the sound of my "orcus" fullrange speakers, I have been tweaking the EQ to get it to sound subjectively good for the music I listen to. But I wanted something a little less subjective.
So, I made a "test" file using Audacity to create a bunch of tones, each one at the respective band that the pulseaudio multiband eq uses.
Then, I tweaked the EQ so that each band had the same subjective loudness, to my ears.
Is this an acceptable method? Or do I need to account for the logarithmic perception of human hearing? If so, is there a simple way of doing this? I'm fairly math-savvy, calculus is not a problem, etc.
Thanks!
Member
Joined 2009
Subjective equalization can work out but measuring is always better.
Some people equalize by ear while listening to frequency sweeps or pink noise.
Honestly, I have trouble using either.
Keep in mind that peaks are more easily heard than dips.
You can try Harman Kardon's How to Listen training software. It helps you develop sensitivity to irregularities in the sound. I'd say spend about an hour with the software (preferably on a good system) and then attempt to do a by ear equalization of your speakers to your favorite music.
I'm not familiar with pulseaudio but it's preferred to use a parametric EQ.
Good luck!
Some people equalize by ear while listening to frequency sweeps or pink noise.
Honestly, I have trouble using either.
Keep in mind that peaks are more easily heard than dips.
You can try Harman Kardon's How to Listen training software. It helps you develop sensitivity to irregularities in the sound. I'd say spend about an hour with the software (preferably on a good system) and then attempt to do a by ear equalization of your speakers to your favorite music.
I'm not familiar with pulseaudio but it's preferred to use a parametric EQ.
Good luck!
Two points:
Firstly, testing with pure tones can give quire misleading results. It would be better to use narrow band noise. You could probably make suitable test signals by playing pink noise through the equalizer with one band set to max boost, and the rest set to max cut.
Secondly, you need to take account of the ears non-flat frequency response. e.g. See here: Fletcher Munson curves - Wikipedia. Also bear in mind that everybody's ears are a bit different. It would be interesting if you could get a few friends to independantly set the eq according to their ears, then compare the settings.
Hmm. After hours of tweaking I am going crazy. I thought I could get a little more objective, but trying to compare my own hearing to the fletcher munson curves would be pointless -- I know for a fact my high-frequency hearing is pretty "shot" from years of shooting guns and loud concerts. Plus, all the equal loudness contour graphs I've found lack the resolution I would need to make fine adjustments based on them.
I think I will just have to accept the "best-sounding-with-the-music-I-listen-to" curves I have established, until I can afford a good measurement mic.
Unless anyone here thinks I could do a decent job with a Shure SM-57.
I think I will just have to accept the "best-sounding-with-the-music-I-listen-to" curves I have established, until I can afford a good measurement mic.
Unless anyone here thinks I could do a decent job with a Shure SM-57.
The fletcher-munson or neebster-cyclecamper curves are not an issue as you are setting to 'subjective' loudness which will take that into account. You can probably learn to EQ this way, and become somewhat reproducible. Lordy knows in the old days we'd set up a huge PA with only ears to guide us.
But really now...the Behringer mic and the Dayton are both cheap. There's articles on making your own for almost nothing (Panasonic capsule on the end of a pencil).
But really now...the Behringer mic and the Dayton are both cheap. There's articles on making your own for almost nothing (Panasonic capsule on the end of a pencil).
Member
Joined 2009
Measurements will tell you when there's something wrong with your system. Sometimes it will be easily correctable by moving the speakers closer or further from the wall.
With a fullrange there's not much to go wrong so doing EQ by ear should be OK. I will not advise doing crossovers without measurements.
Simulation software like WinUSD and Edge can give you very good clues about some things like where the low end on your system falls out (F3-F10) and where the baffle step compensation should be.
If I'm not mistaken the Shure 57 has a very general frequency response graph in the datasheet. You can certainly start from there.
If doing EQ by ear I recommend the following method. Make a peaking filter of 6dB. Listen critically to familiar material and write down conclusions of how this filter affects the sound. Then move the filter to the next octave (100-200-400-800..Hz). Repeat afterwards with a -6dB filter. When you find what frequencies need adjusting you can experiment with different shape and level of filters. Try it on different songs before you settle on something final.
The Munson curve does not apply to equalization!! Unless you listen to all your music at one particular level. Bass will not be heard well at low volume and that's a fact of life. The best solution so far is the way some computer speakers have a desktop knob for the subwoofer. Turn it up when playing quietly, turn it down when it starts flapping.
With a fullrange there's not much to go wrong so doing EQ by ear should be OK. I will not advise doing crossovers without measurements.
Simulation software like WinUSD and Edge can give you very good clues about some things like where the low end on your system falls out (F3-F10) and where the baffle step compensation should be.
If I'm not mistaken the Shure 57 has a very general frequency response graph in the datasheet. You can certainly start from there.
If doing EQ by ear I recommend the following method. Make a peaking filter of 6dB. Listen critically to familiar material and write down conclusions of how this filter affects the sound. Then move the filter to the next octave (100-200-400-800..Hz). Repeat afterwards with a -6dB filter. When you find what frequencies need adjusting you can experiment with different shape and level of filters. Try it on different songs before you settle on something final.
The Munson curve does not apply to equalization!! Unless you listen to all your music at one particular level. Bass will not be heard well at low volume and that's a fact of life. The best solution so far is the way some computer speakers have a desktop knob for the subwoofer. Turn it up when playing quietly, turn it down when it starts flapping.
EQing by ear will drive you crazy... but you can accomplish a few things. If you have generated narrow bands of pink noise that correspond to the sliders on your eq, play each and adjust right vs left for equal loudness. This should result in a centered image for each band. Once all bands match, this will at least have the speakers delivering the same net effect at your listening position. After that, adjust both channels the same amount relative to other bands to make a given range more or less prominent and taste will have to guide you there. I would think you could get things to sound pretty good. Good luck.
Member
Joined 2009
That doesn't sound correct. If by LEFT and RIGHT you mean the two channels of a stereo setup then no, don't pan different frequency bands! Pan only the entire output of the speaker until you get a centered image. The best way is to play both speakers the same loundness and always sit in the middle. Having both speakers at equal distance to the side walls helps a lot too.
It's not a good idea to have different EQ on the left and right.
yes, boris. that's exactly what I mean. if the same sound, same intensity is played by your two speakers, it should sound from your seating position like it is centered between them. Good setup is important as you stated but I would expect that any two speakers will have different environments to fight and will need different treatments to have the same effect on you. if a sound is attenuated by the room for the left speaker (for instance) but not for the right, this can be compensated for and why the eq has separate controls for each channel. just my two cents 
If trying to EQ by ear, be sure to use a reasonably loud volume to minimize the chance that you are correcting for Fletcher-Munsen rather than the speaker's tonal balance.
Myself I have always listened to a pretty well definable SPL; my hand just naturally turns the knob until it sounds right, and a month later if I bother to look, well, there it is ... same O-clock on the dial. I might go louder than that but rarely quieter (assuming I'm doing some listening to music, as compared to "playing music" while occupied on something else). The range is remarkably limited and consistent.
Few people are *totally* deaf near the limits of their hearing, so perhaps you can notice stress or distortion at the upper limit of your hearing that indicates you're compensating too much by overdriving the speaker element at that frequency.
We tend to prefer speakers that are balanced as far as low and high frequency extension goes; if a speaker is bass-shy you will probably prefer the treble rolled off a bit too.
Ideally you would use instrumentation for the "fixed" devices ... amp, speakers, etc ... and then tweak gently by ear. Gently.
You may see a trend that could be compensated for in the Xor design, for example. It is better to make the change in the Xor than to try an equalize for it.
So make such a change, and confirm the results are as expected or desired. Although not everyone agrees that "flat" is the best way to listen, most everyone agrees that the system should should good and tonally balanced flat and at a reasonably loud level. If some tonal adjustment is what you then prefer, then it will be modest, probably.
From that point, it doesn't matter what you use for equalization, since you are only equalizing for your preferences, and a return to "flat" is easily accomplished by just returning to flat on the EQ. You could then compensate for, say Fletcher-Munsen at a low volume but you can always find the base tuning whenever you desire by removing EQ altogether.
I agree that pure tones are not ideal for level setting. Some broadband noise almost always works better. Back in the day we just used to tune the FM tuner off frequency and kill the muting.
Also, when tuning by ear it often helps to just reduce your EQ a bit; this tends to sound better in the long term than what you might decide upon in some intense EQ session, especially if you are a relative novice at EQ by ear. Eg if after a bit of tweaking you find you are ready to quit tweaking and start listening, you have a band +6 dB, drop it to +4. If the adjacent band is +3 dB, drop it to +2. And so on. The more experienced you are at EQ by ear, the less you might have to drop or raise toward the 0dB line, but your corrections will probably be more modest in the first place.
Adjacent bands should be more similar than different so if you have a slider that sticks out as a large correction, maybe tweak it down and move the adjacent ones up a little. If the EQ is parametric you achieve the same thing by adjusting the Q.
Go easy and be smooth on EQ. You generally can't fix a huge aberration with EQ anyway; you should instead look at correcting the underlying problem instead of attempting to EQ it away.
Myself I have always listened to a pretty well definable SPL; my hand just naturally turns the knob until it sounds right, and a month later if I bother to look, well, there it is ... same O-clock on the dial. I might go louder than that but rarely quieter (assuming I'm doing some listening to music, as compared to "playing music" while occupied on something else). The range is remarkably limited and consistent.
Few people are *totally* deaf near the limits of their hearing, so perhaps you can notice stress or distortion at the upper limit of your hearing that indicates you're compensating too much by overdriving the speaker element at that frequency.
We tend to prefer speakers that are balanced as far as low and high frequency extension goes; if a speaker is bass-shy you will probably prefer the treble rolled off a bit too.
Ideally you would use instrumentation for the "fixed" devices ... amp, speakers, etc ... and then tweak gently by ear. Gently.
You may see a trend that could be compensated for in the Xor design, for example. It is better to make the change in the Xor than to try an equalize for it.
So make such a change, and confirm the results are as expected or desired. Although not everyone agrees that "flat" is the best way to listen, most everyone agrees that the system should should good and tonally balanced flat and at a reasonably loud level. If some tonal adjustment is what you then prefer, then it will be modest, probably.
From that point, it doesn't matter what you use for equalization, since you are only equalizing for your preferences, and a return to "flat" is easily accomplished by just returning to flat on the EQ. You could then compensate for, say Fletcher-Munsen at a low volume but you can always find the base tuning whenever you desire by removing EQ altogether.
I agree that pure tones are not ideal for level setting. Some broadband noise almost always works better. Back in the day we just used to tune the FM tuner off frequency and kill the muting.
Also, when tuning by ear it often helps to just reduce your EQ a bit; this tends to sound better in the long term than what you might decide upon in some intense EQ session, especially if you are a relative novice at EQ by ear. Eg if after a bit of tweaking you find you are ready to quit tweaking and start listening, you have a band +6 dB, drop it to +4. If the adjacent band is +3 dB, drop it to +2. And so on. The more experienced you are at EQ by ear, the less you might have to drop or raise toward the 0dB line, but your corrections will probably be more modest in the first place.
Adjacent bands should be more similar than different so if you have a slider that sticks out as a large correction, maybe tweak it down and move the adjacent ones up a little. If the EQ is parametric you achieve the same thing by adjusting the Q.
Go easy and be smooth on EQ. You generally can't fix a huge aberration with EQ anyway; you should instead look at correcting the underlying problem instead of attempting to EQ it away.
Last edited:
Member
Joined 2009
yes, boris. that's exactly what I mean. if the same sound, same intensity is played by your two speakers, it should sound from your seating position like it is centered between them. Good setup is important as you stated but I would expect that any two speakers will have different environments to fight and will need different treatments to have the same effect on you. if a sound is attenuated by the room for the left speaker (for instance) but not for the right, this can be compensated for and why the eq has separate controls for each channel. just my two cents
OK, I see what you are getting at. I'm a bit skeptical, but to be honest I haven't tried it and I can't say if it will work well or not. I'd say it's only an issue if you have very different environments on each side. I have a wall on my right and a big window on my left and there's no significant smearing of the image. The distance to the side walls and the directivity of the speaker seem to be much more important.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.