What is your favorite house curve

[HarmonicTHD]

Will In-ear BINAURAL MICS or headphone fixture such as GRAS 45CA, better than an omni microphone in the room measurement?

No. The cheap UMIK1 is all what is needed. Simply put: It should be omnidirectional to capture the reflections, it also should have a calibration file and lastly the SNR is not so important either for room measurements (better mics ofter have, next to othen characteristics, also better SNR). And there is no mic in the world which is capable of the processing your ears / brain do anyhow. This is what Floyd probably meant, the most difficult part is how the measured room response translates to individual perceptions, as this requires elaborate, controlled preference studies (of which F. Toole conducted a lot).

 

[fpitas]

Measuring is great, and will usually get you close. But some amount of tweaking is usually necessary to get the timbre exactly how you like it.

 

[Toltek]

Green is speaker on-axis response, from 300 Hz and up gated at 4 ms, no smoothing, below 300 - woofer closed miked, baffle step corrected.
Red - speakers at listening position, MMM RTA within 1 ft. Strong room modes peaks and somewhat hot mids
Blue - subjectively best result so far, 2 Subs added and Dirac +4db Bass -3dB treble curve applied.

 

[Floyd Toole]

Will In-ear BINAURAL MICS or headphone fixture such as GRAS 45CA, better than an omni microphone in the room measurement?

No, the brain to process the sound received by the two mics is missing. Using a dummy head mic still sums the direct and all reflected sounds without regard to the direction or time of arrival. Our binaural hearing system pays attention to these differences. Consequently, "room curves" exhibit detailed up-and-down variations caused by acoustical interference that look alarming - and many systems equalize them. However, doing so screws up the direct sound - which is the key factor. You cannot avoid starting with a well designed loudspeaker, or one that is equalized based on anechoic data. Beyond that, above about 500 Hz, one can experiment with broadband "tone control" kinds of EQ, but remember that if this is done "by ear", the program peculiarities are included and these are not constant.

 

[Floyd Toole]

How would a person use the spinorama data to then use EQ to produce a desired house curve for the given speakers? Does that mean a person should not use EQ to change a slight v shaped response curve into a slightly down-tilted sound curve that loses say 10dB from 20Hz to 10KHz?

The "early reflected" curve in a spinorama is a good predictor of the shape of a steady-state room curve above about 500 Hz. Again, a room curve is NOT a target, it is a result, and the result of a well designed forward-firing loudspeaker is a relatively linear, slightly downward tilted steady-state room curve. If the curve you measure is not that shape one must suspect that the loudspeaker has some "issues". This is where only a set of anechoic measurements - like a spinorama - is needed to determine what might be wrong and to determine whether equalization is a possible remedy. Frequency dependent directivity variations are a common problem, and it is the off-axis sound that dominates the shape of steady-state room curves. A spinorama would have revealed the problem from the outset, and perhaps have modified a purchasing decision. Beyond that, one is free to employ broadband humps and tilts to compensate for program variations - a subjective judgement, not a "calibration".

 

[luft262]

The "early reflected" curve in a spinorama is a good predictor of the shape of a steady-state room curve above about 500 Hz. Again, a room curve is NOT a target, it is a result, and the result of a well designed forward-firing loudspeaker is a relatively linear, slightly downward tilted steady-state room curve. If the curve you measure is not that shape one must suspect that the loudspeaker has some "issues". This is where only a set of anechoic measurements - like a spinorama - is needed to determine what might be wrong and to determine whether equalization is a possible remedy. Frequency dependent directivity variations are a common problem, and it is the off-axis sound that dominates the shape of steady-state room curves. A spinorama would have revealed the problem from the outset, and perhaps have modified a purchasing decision. Beyond that, one is free to employ broadband humps and tilts to compensate for program variations - a subjective judgement, not a "calibration".

I'm flattered you responded to my post. You're known to be one of the most important people in audio. I don't want to waste your time so others are free to answer this, but I guess I don't see why it's OK to correct headphones to the Harmon curve using EQ, but apparently not as effective to do something similar with speakers? I understand that below 500Hz it's getting room dependant and above is speaker dependant, but why not compensate for imperfect speaker design with EQ to match a generally accepted response curve?

 

[Floyd Toole]

I'm flattered you responded to my post. You're known to be one of the most important people in audio. I don't want to waste your time so others are free to answer this, but I guess I don't see why it's OK to correct headphones to the Harmon curve using EQ, but apparently not as effective to do something similar with speakers? I understand that below 500Hz it's getting room dependant and above is speaker dependant, but why not compensate for imperfect speaker design with EQ to match a generally accepted response curve?

Headphones deliver sound to the eardrums avoiding all the complications of the three-dimensional sound field in rooms. It is an uncomplicated system, so simple EQ works perfectly. In rooms EQ cannot independently address problems with the direct sound and problems with the reflected sound field - but the human hearing system can. "Imperfect speaker design" takes several forms, not all of which can be addressed by equalization - that is why it is so important to use whatever information you can find and understand - the spinorama being the most useful I can think of - to identify well-designed loudspeakers. A flat and smooth on-axis/listening window response is an essential starting point, but it is the off-axis sound, the "early reflections" revealed in the spinorama, that determine the shape of room curves. These should be smoothish and linear, but not necessarily flat. If that is not the case EQ cannot repair loudspeaker directivity. That done, all else is much easier - fix the bass in your room and use tone controls to adjust spectral balance to compensate for program variations as necessary, and enjoy.

 

[LTig]

I'm flattered you responded to my post. You're known to be one of the most important people in audio. I don't want to waste your time so others are free to answer this, but I guess I don't see why it's OK to correct headphones to the Harmon curve using EQ, but apparently not as effective to do something similar with speakers?

Headphones do not suffer from room reflections so off axis frequency response has no influence on sound - you listen on axis only. Hence one can use EQ to flatten frequency response.

I understand that below 500Hz it's getting room dependant and above is speaker dependant, but why not compensate for imperfect speaker design with EQ to match a generally accepted response curve?

You can use EQ to flatten the on axis frequency response, but if the off axis frequency response is not smooth the resulting room sound isn't smooth either.

 

[luft262]

Headphones do not suffer from room reflections so off axis frequency response has no influence on sound - you listen on axis only. Hence one can use EQ to flatten frequency response.

You can use EQ to flatten the on axis frequency response, but if the off axis frequency response is not smooth the resulting room sound isn't smooth either.

Does this mean if my on axis frequency response, as measured with REW and UMIK-1, is pretty much flat from 10Hz to 10KHz I should not use the "tilt" function, for example, in Audyssey to create a downward slope of 5 to 10dB from 10Hz to 10KHz but instead leave the system alone above 500Hz? Obviously, personal preference comes into play, but essentially I should not assume that a slight downward slope in frequency response is preferable or even achievable with EQ despite what REW is showing me?

 

[HarmonicTHD]

Does this mean if my on axis frequency response, as measured with REW and UMIK-1, is pretty much flat from 10Hz to 10KHz I should not use the "tilt" function, for example, in Audyssey to create a downward slope of 5 to 10dB from 10Hz to 10KHz but instead leave the system alone above 500Hz? Obviously, personal preference comes into play, but essentially I should not assume that a slight downward slope in frequency response is preferable or even achievable with EQ despite what REW is showing me?

Mhhh. Not sure I understand correctly.

The downward slope comes “naturally” when placing an anechoic flat speaker in a room.

Personally I let Audyssey follow that natural downward slope, which has pretty much the same effect as cutting it off above let’s say 500Hz. But this a matter of preference and as stated above use tilt to your liking or program material.

 

[Robbo99999]

Does this mean if my on axis frequency response, as measured with REW and UMIK-1, is pretty much flat from 10Hz to 10KHz I should not use the "tilt" function, for example, in Audyssey to create a downward slope of 5 to 10dB from 10Hz to 10KHz but instead leave the system alone above 500Hz? Obviously, personal preference comes into play, but essentially I should not assume that a slight downward slope in frequency response is preferable or even achievable with EQ despite what REW is showing me?

When you try to measure your on axis frequency response in your room at your listening position you're not measuring just the on axis frequency response of the speaker, but instead all of it's in-room reflections that start off as "rays" that emanate from the speaker at various arced angles (not just on-axis). Besides, terms like on axis frequency response and listening window are terms related to anechoic measurements of speakers (no reflections, no room interactions) - so the advice to buy or EQ a speaker to a smooth listening window with flat on-axis response is based around the anechoic measurements, not your in-room measurements using your UMIK - it's not the same thing.

You'd get the spinorama data from Amir's reviews of the speaker you've decided to buy, and then you'd optimise it with EQ by EQ'ing it to a smooth Listening Window and flat on axis frequency response. So in REW you'd use EQ filters to smooth out the Listening Window response, then you'd apply that same set of filters to the on axis frequency response in REW to see how those filters affect the on-axis response, then you'd tweak that same set of EQ filters to try to get the on-axis frequency response to be flat in tonality (so you'd use broad acting filters at this stage). For example here is the Anechoic EQ I did for my JBL 308p Mkii speakers, (Amir has reviewed this model of speaker, and I used his spinorama data in REW to do the EQ's):

JBL 308P MKII Studio Monitor Review

Just picked up a pair of these to play with while in COVID QT and have a question related to the limiter. I was hitting one of the speakers pretty hard (didn't measure the actual input voltage or output SPL) to see if I could duplicate the limiter symptoms Amir got...which I thought I did. I...

www.audiosciencereview.com

Notice in the graphs at that link that I've prioritised smoothing out the Listening Window (removing the "jaggies" and larger undulations/deviations), and notice that in the On-Axis graph that I've prioritised making sure the tonality is totally flat, you can see it tracking 109.6dB on average as flat along that line as possible. It's a juggling act between smoothing out the Listening Window response whilst making sure that the same set of filters are working to make the On-Axis Response track a flat dB line (horizontal flat). Thankfully the JBL 308p Mkii is a very well behaved speaker in it's directivity and frequency response which means that it's an easy juggling act between the Listening Window and the On-Axis.

 

[luft262]

When you try to measure your on axis frequency response in your room at your listening position you're not measuring just the on axis frequency response of the speaker, but instead all of it's in-room reflections that start off as "rays" that emanate from the speaker at various arced angles (not just on-axis). Besides, terms like on axis frequency response and listening window are terms related to anechoic measurements of speakers (no reflections, no room interactions) - so the advice to buy or EQ a speaker to a smooth listening window with flat on-axis response is based around the anechoic measurements, not your in-room measurements using your UMIK - it's not the same thing.

You'd get the spinorama data from Amir's reviews of the speaker you've decided to buy, and then you'd optimise it with EQ by EQ'ing it to a smooth Listening Window and flat on axis frequency response. So in REW you'd use EQ filters to smooth out the Listening Window response, then you'd apply that same set of filters to the on axis frequency response in REW to see how those filters affect the on-axis response, then you'd tweak that same set of EQ filters to try to get the on-axis frequency response to be flat in tonality (so you'd use broad acting filters at this stage). For example here is the Anechoic EQ I did for my JBL 308p Mkii speakers, (Amir has reviewed this model of speaker, and I used his spinorama data in REW to do the EQ's):

JBL 308P MKII Studio Monitor Review

Just picked up a pair of these to play with while in COVID QT and have a question related to the limiter. I was hitting one of the speakers pretty hard (didn't measure the actual input voltage or output SPL) to see if I could duplicate the limiter symptoms Amir got...which I thought I did. I...

www.audiosciencereview.com

Notice in the graphs at that link that I've prioritised smoothing out the Listening Window (removing the "jaggies" and larger undulations/deviations), and notice that in the On-Axis graph that I've prioritised making sure the tonality is totally flat, you can see it tracking 109.6dB on average as flat along that line as possible. It's a juggling act between smoothing out the Listening Window response whilst making sure that the same set of filters are working to make the On-Axis Response track a flat dB line (horizontal flat). Thankfully the JBL 308p Mkii is a very well behaved speaker in it's directivity and frequency response which means that it's an easy juggling act between the Listening Window and the On-Axis.

So basically, if I can find the spinorama for my speaker, such as in a review from Amir, I can just plug in the PEQ's that he used to correct the speaker, even though my room is different, because above 500Hz it's the speaker and not the room. After that tilt and such is just preference? What if Amir used a PEQ filter on the speaker below 500Hz? Should I disregard that since it would depend on the room? Is there a step by step guide somewhere that shows how to do this? Or a video?

 

[ernestcarl]

I

So basically, if I can find the spinorama for my speaker, such as in a review from Amir, I can just plug in the PEQ's that he used to correct the speaker, even though my room is different, because above 500Hz it's the speaker and not the room. After that tilt and such is just preference? What if Amir used a PEQ filter on the speaker below 500Hz? Should I disregard that since it would depend on the room? Is there a step by step guide somewhere that shows how to do this? Or a video?

Feel free to download any published speaker data by Amir to try out (ones you are mostly interested in). Open or copy the files to a spreadsheet program like Excel. Select the columns of data you want to import into REW (ones not already separated) and paste each set to a new text file labeled LW, ER, PIR, etc. You only need the frequency and amplitude information. Drag or open these files into REW and modify the curves using the EQ window. If you want to view more than one “predicted SPL” curve at a time, use the “overlays” window.

I myself did not copy verbatim the PEQs used by Amir or the ones generated by Maiky. I measured and listened to my speakers using their provided filters. I thought Amir’s HF shelving was unnecessary (made the speaker sound dull) and Maiky’s filters tried to flatten the response too much making the sound too bass light. As Maiky himself writes: test the filters out for yourself if you like how they sound or not. Also compare the results to your own in-room measurements. EQ requirements in the bass may very well vary from one room to the next.

 

[Tom C]

The most important determinant of sound quality is: wait for it. Pay attention now. Ears open. Ready? Here it is: Direct sound! Yes, direct sound trumps room reflections with respect to human perception of sound quality.
1.) Start with a speaker that gives flat FR. Use anechoic data to select for purchase a speaker with flat FR. What‘s that you say? You already bought a speaker, and anechoic FR isn’t flat? If the speaker you have does not deliver a flat FR in an anechoic chamber, then the best you can hope for is to try flatten, using EQ, based on measurements obtained in anechoic chamber. As close to flat as you can get. How? What guide? Use curve from anechoic measurement to EQ. Cut anechoic peaks. Boost anechoic dips. Mind the Q’s.
2.) Select for purchase speaker with smooth off-axis behavior, based on anechoic measurements. Now here, if your speaker does not have smooth off-axis behavior, you’re SOL. Because nothing, not even the most excellent room treatment, will pull your bacon out of the fire. Nothing can fix it. Best use poor off-axis speakers for a pair of headphones, because if you jiggle your head even a little bit, you’ve altered what hits your ears, to the point that FR is not smooth any more. These speakers can only give one tight little sweet spot where FR is not bumpy (same as saying smooth), and that is directly on-axis. Don’t sit too far away, and don’t move your head, because if you do, the room takes over. Well, there is other one fix: buy better speakers.

So, if my speaker‘s on-axis Klippel line is flat, and the DI is smooth, I’m golden? Yes! Physics will make such a speaker’s actual output in your room conform to a downward sloping curve. Which is, really, only what it looks like to your eye. What you’re truly after is not so much a treble cut as a bass boost. But boosting bass has its own challenges that are best dealt with separately. Because, if there’s anything I got out of calculus, it’s that a complex problem is best dealt with by breaking it into manageable parts.

 

[dlaloum]

Question - is the measurement curve shown by the Dirac software, the "Room" curve (ie: slow/continuous measurements of the direct as well as the reflected sound averages) - or the "Direct" signal (typically measured by gating it around the first 4ms or similar)?

(This may be a question best asked on the Dirac thread.... )

 

[Lexxie]

Hmmm, this thread is still about best EQ or FR for House Music? I love HM/EDM/etc., but ...

At the risk of being obvious, these tracks are so mixed, EQ'd, produced already, that they have strong "artificial profile" injected into them. In no other genre is the need for manually fiddling with your own EQ as necessary as in this genre. Or, not fiddling... So, probably I think this question can't be answered so well. Except to say they generally do a lot of "U" or "V" in the recording of it, such that, somewhat counterintuitively, I "trust" their own colour they added to it and like to go to a semi-flat profile similar to Classical music. Results may vary by the recording (and often do!)

 

[Dj7675]

Over the past quite a few years I have tried to get the perfect room curve and force speakers to that curve. I successfully was able to get the curve to match various house curves, but always ended up finding the sound lacking in the long run. In my room, well designed speakers actually measure fairly flat from 300hz to 10khz. When I force a 10dB drop from 20hz-20khz, it doesn’t sound very good. This was with a variety of very good speakers from Revel, JBL and others where Spinorama showed they were well designed speakers. After reading THIS post by @thewas that compiled comments by @Floyd Toole I was thinking of things backwards. I didn’t need to force a room curve on a good, well designed speaker. For the most part the work was done. If I had Spinorama I could create a few filters to attempt to fix an issue with manual EQ and turn the filter on/off to see if it made a positive difference or not when downmixing to mono. We are lucky to have people like @amirm as well as Pierre and his site https://www.spinorama.org/ to provide speaker info. When I do full range automated EQ, and have a good speaker, I am not much more careful to follow their natural response, although I still haven’t had full range automated EQ sound as good as limited with a few filters over around 500hz.
I guess my point is my experience is more in line with comments by @Floyd Toole regarding expecting any predefined, automatically corrected room curve to necessarily sound as good as the graph looks.

 

[Robbo99999]

So basically, if I can find the spinorama for my speaker, such as in a review from Amir, I can just plug in the PEQ's that he used to correct the speaker, even though my room is different, because above 500Hz it's the speaker and not the room. After that tilt and such is just preference? What if Amir used a PEQ filter on the speaker below 500Hz? Should I disregard that since it would depend on the room? Is there a step by step guide somewhere that shows how to do this? Or a video?

@ernestcarl gave you a good response there. If you've got enough PEQ filters to play with I'd correct the response below 500Hz too before you start any in-room measurements. You'd then do roomEQ on top of that. You could tilt the entire curve (from 20Hz-20kHz) linearly using broad Tone Control after that on a track or per albumn basis as sometimes some artists record a bit too bright or too dull - but to be honest I've found that an Anechoic EQ combined with roomEQ is correct tonality on almost everything (I just leave it, I don't do Tone Control on a per track or artist basis).

 

[luft262]

So, to summarize what I'm hearing/learning...and please correct me if I'm wrong, the general process would be:

1. Use spinorama to get the most accurate speakers you can afford (because EQ will not be able to fix everything, especially off-axis sound)
2. Position your speakers/subs optimally in the room
3. Use EQ (automated, manual, or both) below roughly 500Hz to flatten out the response curve

Below this step is where I'm getting confused

4. Above 500Hz should I be using EQ on the speakers based on what I measure myself in the room or what is visible in the anechoic spinorama data?

5. Increase/decrease bass and/or adjust tilt/tonality as desired

Thank you all for your sage advice.

 

[ernestcarl]

4. Above 500Hz should I be using EQ on the speakers based on what I measure myself in the room or what is visible in the anechoic spinorama data?

I'd say both -- the former being highly contingent on the reliability of your own measurement skill and ability to interpret the data. Professional acousticians and calibrators do it all the time. Also, there is such a thing as unit to unit variation... For example, as far as I know, and as stated in one of Amir's speaker reviews there are two versions of the Fluid FX50 with slightly different DSP tunings. I know my own orignal JBL LSR305s do not match each other -- the differences are small in the grand of scheme of things -- but other ASR members have reported the similar issues. And what is the speaker configuration in the room? My FX50 is currently elevated above my bedroom TV and pointing at the window above my head, so, of course, I do not care on how it sounds on-axis but rather off-axis while I'm sitting or lying on the bed.