I'm trying to eq a set of bookshelf speakers and every time I measure them I get slightly different results, I'm not sure where I'm going wrong. I'm using rew to measure and a umik-1 with a calibration file. What I do is use a chair and a few textbooks to prop the mic up to listening level about three feet away from one of the speakers, and unplug the other channel. I make sure to have the mic set to left if I'm measuring the left channel or right if I'm measuring the right. Is there anything more to it? For example, yesterday it sounded to me like my speakers had a hump in the midbass but the graph didn't show it, it showed it as a dip for some reason. The next day I measured again and it showed up on the graph, confirming what I heard. I really don't understand this and I want to figure it out soon so I can eq the speakers.
IF you are trying to EQ them into the room, for best sound at your listening position, you need to set your mic up at about ear level on your listening position... Take a sweep for L+R and look at your response curve.
Try to get your readings at about 1/2 power... speakers are the most honest when not being driven hard. (I typically work at 50 or 60 db levels)
From there you can either manually set the EQ in your amp or if you are using a DSP unit REW recognizes you can ask it to create filters for you.
And no... no two scans will ever return exactly the same result... the air pressure changed, you moved your coffee mug, you stood in a different spot, the humidity changed, the volume wasn't set exactly as before... and on and on and on... They will be close, within a couple of db, but never exactly the same.
Also a little hint when setting your EQ or DSP filters... don't try to get an exactly straight line from REW... I've only seen that once and it was a total fluke. If you get within plus or minus 3 db you can call it mission accomplished and go hug your spouse.
Last edited:
Buy yourself a proper mic stand, your umik cost good money (I use the same) its a shame to compromise it.
Mic stand a good idea. Do three runs about a foot apart around your chair and average in REW.
Do L and R separately. When done together, there are always odd interactions (phase interference) in the treble when running sine waves that are irrelevant to stereo music playing. And the EQ will be different for L and R.
Otherwise, good advice from Douglas Blake.
Actually, Step One is to set up system to taste by ear. Then see what the plots look like. Then touch-up the EQ. People who think that's backwards are people who would buy a meter to tell 'em if they're hungry.
B.
Last edited:
Hello Ben,
Good advice. Thanks for chiming in.
I don't generally use separate left and right scans, unless there is something pointing me to a wild difference between the speakers... like sound displacement or an audible room echo. I like to keep it simple so that my friends/clients can do their own touch-ups and personalisation later.
I do the instrument setup first... personalizations later. It's been my experience that left to their own devices most people will paint that little smiley face on their frequency curves and then wonder where the vocals went.
... and don't be getting between me and a good meal!
Good advice. Thanks for chiming in.
I don't generally use separate left and right scans, unless there is something pointing me to a wild difference between the speakers... like sound displacement or an audible room echo. I like to keep it simple so that my friends/clients can do their own touch-ups and personalisation later.
I do the instrument setup first... personalizations later. It's been my experience that left to their own devices most people will paint that little smiley face on their frequency curves and then wonder where the vocals went.
... and don't be getting between me and a good meal!
Unless your system is absolutely perfect head to toe and your chair is perfectly equal distance, there will be cancellations somewhere on the compass on pure sine waves in the treble where the wavelengths are small. But not relevant to stereo music where that kind of math doesn't apply.
Got to be the lucky room (and lucky spousal approval) where L and R plot the same.
If your system is nicely matched, as you say, then why bother with both speakers when just L will do?
Easy to eyeball: just run the three curves, L, R, and both. Where "both" is lower than L or R, you've got cancellation.
Whether doing FR or ear first, important to resist the wannabee engineer false belief that flat FR is "correct" sound.
B.
Typically I set them up by measurement within a 1/8 of an inch. Toe-in is set using laser pointers on the sweet spot.
Please, don't hand me this garbage from some guy's book to deal with.
Simple answer ... I don't buy it for a second.
There really are two questions (broadly speaking) and the first is - are the speakers producing a smooth response before the influence of the room. You're measuring at 3 feet, this is good. You want to pull the setup away from walls to keep this clean.. the further away from any surface, the lower you can cleanly measure in frequency. Don't expect to trust the measurement in the bass frequencies, the room walls are too close for that. You want to measure with a gated sine sweep. The gating is essential to remove the reflections.
As an actual engineer, I do find that flat frequency response sounds right to me.
I have the capacity to add arbitrary EQ to my HiFi setup while the signal is still in the digital domain, so I can experiment as much as I like. Having tried a lot of curves, flat is the one I keep coming back to.
Occasionally, I use my HiFi system to mix and master records, and I want it to be flat then, too.
Chris
I can EQ almost anything any way I want in my HTPC using the Equalizer APO which provides a rich selection of filters and even a 32 band graphic equaliser. So, yes, I've experimented with lots of different "curves"... and I always end up resetting everything to flat... Not electronically flat, acoustically flat in my room.
So lets just take a step back and acknowledge the elephant in the room...
What are you listening to... the music, or your system?
The whole concept behind "High Fidelity" is that a home system should reproduce sound that is true to the original. Of course we know this isn't going to happen once the recording, mixing and mastering engineers get their hands on it... so the best we can hope for, in our listening areas, is sound that is true to the source recording we are listening to. We need to also understand that these recordings are highly variable, mixed for artistic, not technical, satisfaction.
To achieve this, we need a system that does not exaggerate or deplete anything within the agreed response range of 20hz to 20khz. Of course this too is not going to happen. But with modern gear we can get "close as makes no difference" which usually amounts to plus or minus 3db of the ideally straight line. (acoustically, in the room)
It is only in this state that our systems step aside and let us hear what is recorded. Be it good or bad, it is reasonably accurate.
The minute you introduce a deliberate curve into that equation, you are doing two things... you are exaggerating some part of the recorded sound and you are introducing distortion of the frequency response of your listening room.
For example, if you are using a curve that boosts the bass by +6db (which is not uncommon, btw) and your recording is boosting the bass by +6db, now you have a very audible +12db exaggeration of the sound... and that, my friends is not what HiFi is about.
We seem to forget that we are all slaves to our sources. Nothing we do to our systems will fix a bad recording... no matter how hard we try. But, there are a lot of things we can do to make a perfectly good recording sound terrible.
==
Now, on the other hand... if you are listening to your system, your speakers... Go ahead, do whatever the heck you like. Just don't try to sell it as "correct", "better" or "more realistic"... it's none of that.
Last edited:
That's all well and good if you are modifying the speakers.
Tuning speakers into a room is an entirely different process, Allen.
You've done this argument before, and it still does not make sense.
Firstly you state that you want to hear the recording 'as intended by the producer'
Therefore whatever goes onto the CD / released track must be regarded as 'flat' (the recording cannot be boosting bass by +6dB as a CD is to be regarded as a 'flat' part of the chain/ how the producer intended.)
If the recording is played back on a system with +6dB of bass boost then you will have that +6dB of boost, not the 12dB you keep alluding to.
Yep... compared to the recorded media, that is correct.
However, if the recording is of a live performance, such as classical or jazz, then the double bump does happen.
Ummm, I seem to have aroused the ire of the wannabee engineers. To state their error in a single phrase, they confuse FR with human hearing.
Even if you favour a non-sensical undefinable "...as if you were there....." criterion, two speakers can not put you "there" any more acoustically than physically. There is simply no way to physically specify the sound in your room at home that situates your ears in the jazz club. But some recordings can played on some systems at some volume levels to give a feeling reminiscent of the club.
As far as taste in FRs, it may be that FR-flat sounds right if you listen to gawd-dawful pop recordings with their exaggerated bass levels and/or you consider that sound to be your benchmark.
B.
Even if you favour a non-sensical undefinable "...as if you were there....." criterion, two speakers can not put you "there" any more acoustically than physically. There is simply no way to physically specify the sound in your room at home that situates your ears in the jazz club. But some recordings can played on some systems at some volume levels to give a feeling reminiscent of the club.
As far as taste in FRs, it may be that FR-flat sounds right if you listen to gawd-dawful pop recordings with their exaggerated bass levels and/or you consider that sound to be your benchmark.
B.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.