Harman Target Curve

[srm51555]

I've been reading more on implementing room EQ with my DIY M2's and was wondering if anybody has used the Harman Target Curve as a reference?

74876

Sean Olive wrote an article about different Room Correction products here (http://seanolive.blogspot.com/2009/11/subjective-and-objective-evaluation-of.html) back in 2009 that I recently read and it gives two examples where room eq either didn't help or sounded worse than no eq.

Greg Timbers as also noted room correction here (http://www.audioheritage.org/vbulletin/showthread.php?35841-M2-Master-Reference-Monitor-Purchase&p=363503&viewfull=1#post363503), I'm guessing he assisted on the above curve. He mentioned not to mess around with eq above 400hz unless you know about the loudspeaker directivity behavior. Measurements on each side of the two listening positions (around 3.5 feet or 1 meter) I have in my listening room resulted in very good directivity with the M2 Waveguide.

[Don C]

I have something that looks a lot like that dialed into my system. I didn't try to match that curve though, I just set it up by ear. The REW software will suggest a rolloff at higher frequencies that looks similar.

[srm51555]

Looking at the HF roll off of the Harman curve, I was kind of surprised it started at 2.5KHz. Under REW I set a LF target setting starting at 160Hz and set the rise slope of 2.5dB/octave. For HF I set a HF Fall starting at 2.5kHz and a falling slope of 2.5dB/octave. All this was done on the left speaker so I could use the right side for reference. The results were ok, but I'm still new at this so it's probably something I'm doing. Good to know the curve or something similar is used by others here, Thanks for posting.

Thanks,
Scott

[Ed Zeppeli]

sub'd.

[Mitchco]

The Measurement and Calibration of Sound Reproducing Systems (http://www.aes.org/e-lib/browse.cfm?elib=17839) is Tooles paper and includes Olives research. This AES Open Access document can be downloaded without an AES membership. Jumping to Figure 14:


http://i1217.photobucket.com/albums/dd381/mitchatola/toole_zps60mrcuow.jpg

My JBL 4722 stereo config implements a small variation of the trained listeners preferred response above. Flat to 1kHz and then a straight line to -6 to -8 dB at 20 kHz depending on how lively or damped ones room is. And how much high frequency energy one can take with dynamically compressed music.

I agree with Greg Timbers, about eq and directivity. I use full range FIR filter eq, but because of the controlled or constant directivity waveguide, eq'ing the response on axis has a predictable and smooth off axis response across a wide sweet spot. If it is ok to link to another site with a few measurements: http://www.avsforum.com/forum/89-speakers/1925586-just-purchased-pair-jbl-4722n-speakers-198.html#post48767385

Frequency Dependent Windowing (FDW) is the common theme of DSP software like Audiolense, Acourate, Dirac, Denis' DRC RePhase, and others. The idea is that the analysis and correction windows is long at low frequencies and sliding shorter as frequency increases: e.g. 750ms @ 20 Hz, 100ms @ 100Hz, 15ms at 1 kHz, and less than a millisecond @ 20 kHz. This more closely represents what we hear in small room acoustics. Acoustic and Psychoacoustic Issues in Room Correction by JJ Johnston explains why we hear what we hear in small room acoustics, hence why FDW.

So there is an element of the room at low frequencies, in addition to the speaker being equalized and as frequency increases the amount of room sound is windowed out to the point of just shaping the direct sound to match a target response for neutral, accurate sound reproduction. Most of these software packages allows one to custom design a FIR filter, including adjusting FDW size. Denis's open source DRC program is a good software and provides a detailed explanation of FDW, target curves, preringing, etc: http://drc-fir.sourceforge.net/doc/drc.html

Personally, I use Acourate and very happy with the design and sound quality of the FIR filters. The linear phase digital XO is set to the factory recommended frequency of 650Hz for biamp operation and matching directivity of the low and high frequency sections. Digital XO enables one to time align the woofers and tweeters to an ideal speakers step response. The frequency response of the 4722 can be equalized to almost +-1 dB across its operating range, covering a wide sweet spot. As an ex recording/mixing engineer, this is the level of precision and accuracy I look for in a control room/monitor system. But with the high efficiency of the 4722, the sound is dynamic, even at low volumes. I listen mostly at 83 dB SPL which means low distortion from the 4722. It has been fun and sounds great to my ears. Recommended.

Hope that helps.

Cheers, Mitch

[srm51555]

Thanks for the write up, the Toole paper is something on my to read list. I played around more with the curve implementation last night and received much better results. My downward slope was around 14dB with frequencies between ~160Hz and 2.5kHz being flat. I'd like to try a slope of 11-12dB in the coming days.

[srm51555]

I took the Harman Target Curve snapshot posted earlier, added grid lines for finding in between dB values and charted out the following in Excel:

75023
I haven't tried it out just yet, but hope to this Thursday. Column B shows values plotted from Target Curve and Column D is the sum of B and 100 to get a target curve of around 75dB for REW.

Any thoughts?

Thanks,
Scott

[bubbleboy76]

Nice.
Can you please post a REW-house curve file with that column D data?

[srm51555]

Please see the attached 75054.

Here is what REW shows when entering the Target Curve file.



Thanks,
Scott

[srm51555]

I made slight adjustment of the file submitted earlier today. Please see previous post for updated file

[srm51555]

The above file was tried last night and I found it to have too much bass in the lower regions. It may work in other rooms but it's back to the drawing board for me.

Thanks,
Scott

[bubbleboy76]

How did you measure? Single point?

[srm51555]

For this trial, yes. I'm the only one who listens in that spot so I figured it wouldn't matter. I'm currently researching the JBL Synthesis curve which appears to have 2dB less of a rise from 160hz to 40Hz compared the Harman Target Curve. I haven't found a clean picture to plot out the curve like I did with the Harman Curve yet. I plan on trying the Bruel and Kajer Curve as well. I figure all this is good practice for when I eventually try the 2450-SL/Be's.

Thanks,
Scott

[Jonas_h]

The target curve you have, does not look like the Harman curve... Below is the Synthesis target:

75072

[ivica]

The Measurement and Calibration of Sound Reproducing Systems (http://www.aes.org/e-lib/browse.cfm?elib=17839) is Tooles paper and includes Olives research. This AES Open Access document can be downloaded without an AES membership. Jumping to Figure 14:


http://i1217.photobucket.com/albums/dd381/mitchatola/toole_zps60mrcuow.jpg

My JBL 4722 stereo config implements a small variation of the trained listeners preferred response above. Flat to 1kHz and then a straight line to -6 to -8 dB at 20 kHz depending on how lively or damped ones room is. And how much high frequency energy one can take with dynamically compressed music.

I agree with Greg Timbers, about eq and directivity. I use full range FIR filter eq, but because of the controlled or constant directivity waveguide, eq'ing the response on axis has a predictable and smooth off axis response across a wide sweet spot. If it is ok to link to another site with a few measurements: http://www.avsforum.com/forum/89-speakers/1925586-just-purchased-pair-jbl-4722n-speakers-198.html#post48767385

Frequency Dependent Windowing (FDW) is the common theme of DSP software like Audiolense, Acourate, Dirac, Denis' DRC RePhase, and others. The idea is that the analysis and correction windows is long at low frequencies and sliding shorter as frequency increases: e.g. 750ms @ 20 Hz, 100ms @ 100Hz, 15ms at 1 kHz, and less than a millisecond @ 20 kHz. This more closely represents what we hear in small room acoustics. Acoustic and Psychoacoustic Issues in Room Correction by JJ Johnston explains why we hear what we hear in small room acoustics, hence why FDW.

So there is an element of the room at low frequencies, in addition to the speaker being equalized and as frequency increases the amount of room sound is windowed out to the point of just shaping the direct sound to match a target response for neutral, accurate sound reproduction. Most of these software packages allows one to custom design a FIR filter, including adjusting FDW size. Denis's open source DRC program is a good software and provides a detailed explanation of FDW, target curves, preringing, etc: http://drc-fir.sourceforge.net/doc/drc.html

Personally, I use Acourate and very happy with the design and sound quality of the FIR filters. The linear phase digital XO is set to the factory recommended frequency of 650Hz for biamp operation and matching directivity of the low and high frequency sections. Digital XO enables one to time align the woofers and tweeters to an ideal speakers step response. The frequency response of the 4722 can be equalized to almost +-1 dB across its operating range, covering a wide sweet spot. As an ex recording/mixing engineer, this is the level of precision and accuracy I look for in a control room/monitor system. But with the high efficiency of the 4722, the sound is dynamic, even at low volumes. I listen mostly at 83 dB SPL which means low distortion from the 4722. It has been fun and sounds great to my ears. Recommended.

Hope that helps.

Cheers, Mitch

Hi Mitch,
thank You for the info,

regards
ivica

[Mitchco]

Cheers ivica! Scott how goes it? Just curious, did you try flat to 1 kHz, and using 1 kHz as a hinge point, straight line to -6 to -8 dB at 20 kHz?

Kind regards, Mitch

[srm51555]

The target curve you have, does not look like the Harman curve...

I'm beginning to think the Harman Target Curve Bar graph shown in post one is for headphones. Below is the Harman Target Curve Bar Graph in REW with a better windowing of 105/55 to match the Synthesis curve you posted. It shows most of the characteristics of the bar graph but just didn't sound right. The 25Hz marker was left in for reference.

75199


Cheers ivica! Scott how goes it? Just curious, did you try flat to 1 kHz, and using 1 kHz as a hinge point, straight line to -6 to -8 dB at 20 kHz?

Kind regards, Mitch

I will be, most likely tomorrow or Thursday. I will report back my findings.

[srm51555]

I had some time last night to implement a flat to 1kHz then 8dB slope to 20kHz. I popped in my standard test CD of Studio Tan by Frank Zappa and listened for about ten minutes on low volume since it was late. I still need more listening time with these setting but feel these are the best results I have gotten so far. Thanks for the suggestion.

Thanks,
Scott

[bubbleboy76]

The target curve you have, does not look like the Harman curve... Below is the Synthesis target:

75072

Do you have these curves to share, in REW-format, or something like that?