I'm designing a crossover for a 3-way and initially I was targeting a flat response and have evaluated several designs. But then I started thinking about alternative voicings, and then recalled that the human ear has an auditory sensory function. So, if I design for truly flat response, the human ear would, for example, actually perceive 2k-5k much more than other frequencies, which may sound unpleasant. Now, if I design for the human sensitivity function then it should "apparently" sound flat to an ear. Now, my question is, do sound engineers already bake in the human auditory function into their mix?
All well supported arguments appreciated.
All well supported arguments appreciated.
Google Harman Kardon dB curve or similar, it should drop slightly above 2000 hz to be really comfortable
The Fletcher–Munson curves apply to live music the same as recorded music, so the only reason you may want a non-flat response is for "loudness", which is what the "loudness" function on a volume control does. Using a fixed EQ regardless of the volume would not be useful.
https://en.wikipedia.org/wiki/Equal-loudness_contour
https://en.wikipedia.org/wiki/Equal-loudness_contour
Actually, loudness knobs often overdo it.
The required correction depends on the difference (in dB) between the equal loudness contours at the volume at which you listen and at the volume at which the recording was recorded, mixed or mastered. You should therefore get a flat response when you listen at a realistic volume for the genre you listen to, but more often than not, a loudness control only gives a flat response when the volume is very loud. That's fine for music that is supposed to be very loud, but not for softer music.
The required correction depends on the difference (in dB) between the equal loudness contours at the volume at which you listen and at the volume at which the recording was recorded, mixed or mastered. You should therefore get a flat response when you listen at a realistic volume for the genre you listen to, but more often than not, a loudness control only gives a flat response when the volume is very loud. That's fine for music that is supposed to be very loud, but not for softer music.
My "one Old Franc in the bastringue" ... : Speakers and HiFi Systems are they -or not- supposed to give a "live concert" like response, that mean a flat one ? The need of tone adjustement on high freq may be a need as a paliative vs "old sound engineers very high frequencies deaf", obvious on certain records...
Given that the sound engineer monitors the mix on studio monitors with (hopefully) a flat frequency response, the closest you can get to the sound as heard in the studio is to listen on speakers with a flat frequency response.
Short answer: Yes
Long answer: It is important to be true to how you listen to music and movies. Instead of attempting artificial absolutes think about how you actually listen most of the time. If 90% of your listening happens with music in the background or at low volumes why on earth should you pick a speaker by stress testing the dynamic range??
Dali makes a line of speakers with the tweeter significantly elevated. Dynaudio has at least two curves they tune speakers to, and why not?
Long answer: It is important to be true to how you listen to music and movies. Instead of attempting artificial absolutes think about how you actually listen most of the time. If 90% of your listening happens with music in the background or at low volumes why on earth should you pick a speaker by stress testing the dynamic range??
Dali makes a line of speakers with the tweeter significantly elevated. Dynaudio has at least two curves they tune speakers to, and why not?
Interesting point nior69 what is the frequency response of a drum kitt or a tuba or a trumpet.would it be rated + or - 3db. If not would it be audiofile. Home stereo reproducing music is a huge mix of science ,art and consumer pref. Most sound engineers kill the real live sound of music through processing. So build your speaker using your own ears.only you will like or dislike the finished sound.
The question is a little confusing. A passive crossover can't adjust response dynamically to compensate for level. You get one shot. Also, everything you hear, reproduced or live, is identically altered by the structure of your auditory system (HRTF) prior to the stimulus reaching your eardrum. A speaker that emulates your HRTF means doubling up that alteration, once by the speaker and once by your head.
That's logical.. however it's not that critical when a speaker is done right.
Many speakers have significant level dependent issues of the non-linear perception type which give rise to concerns in this area.. but which are actually speaker design problems.
Many speakers have significant level dependent issues of the non-linear perception type which give rise to concerns in this area.. but which are actually speaker design problems.
"Well, akchually..." and dare I mention it ... incandescent light bulbs in series with drivers could do this. I've heard of a light bulb in series with a tweeter as burnout protection. Light bulbs are near-ideal in that their resistance is nearly constant at audio frequencies and so won't add harmonic distortion (except lower bass), but vary with current variations of about a tenth of a second or less. I can imagine a response/volume curve that approximates the Fletcher-Munson curve, though it would be a rather complex crossover to involve different lamps as well as different drivers.
This is somewhat of a wild idea and I really shouldn't send this post, but whatever.
I presume you mean (perhaps among other things) dynamic compression due to voice coil heating?
I like the fact that this being discussed. Right now, I am doing a bit of compensation not to get a flat response (yes, it is one priority), but one that better matches my aging ears. I thought that my speakers were pretty heavy on the bottom, but then realized that I was light on the highs, mids and top. And the beat goes on. Just trying to get a few more good years out of music before it is all gone.
I tend to listen at a relatively consistent level and my system seems to sound the most natural and realistic with a slight tilt to the frequency response, at least measured in-room. The bass is a couple of dB louder than the treble.
I listen to quite a bit of live Classical music, mostly Chamber Orchestra stuff, and that's what I use to tune my system. The more realistic my system is with acoustic music that I have a real frame of reference to, the "better" it sounds with everything else.
Pete
I listen to quite a bit of live Classical music, mostly Chamber Orchestra stuff, and that's what I use to tune my system. The more realistic my system is with acoustic music that I have a real frame of reference to, the "better" it sounds with everything else.
Pete
In my opinion you should definitely voice the speaker to your liking. I mean, you built it for your listening.
Assuming you haven't done a bunch of crossovers (sorry if this is incorrect), don't go into the process expecting the first try to be everything you were hoping for. Give yourself options and try them. Alligator clips are great for crossover development.
Assuming you haven't done a bunch of crossovers (sorry if this is incorrect), don't go into the process expecting the first try to be everything you were hoping for. Give yourself options and try them. Alligator clips are great for crossover development.
There are real reasons not to make a speaker flat.
A positive example is room power, not all speakers are equal in their dispersion.
A negative example, compensating for design problems, diffractions, room reflections etc.
A positive example is room power, not all speakers are equal in their dispersion.
A negative example, compensating for design problems, diffractions, room reflections etc.
In broad terms, targeting a flat on-axis SPL response seems to be the accepted engineering approach to the problem of sound reproduction. Flat response in loudspeakers, amplifiers, CD players, tape decks, etc, is what should be aimed for. If it isn't achieved, a coloured sound signature results, which may or may not be pleasing, dependent of course on the program material and the room acoustics.
If a non-flat response is being considered, just add a graphic/parametric or other type of equalizer into the system and adjust to taste. Of course, the degree of adjustment will depend on the program material, the replay sound pressure level, the room acoustics, etc. Baking anything like that into a loudspeaker seems counterproductive.
Consider an argument for not having a non-flat response. What would be the ideal "voicing" if a sound engineer recorded a set of one-third octave bands of pink noise, one at a time, using a monitor with a flat response to adjust the level of each band? In an equivalent room environment and listening position, what response should my loudspeaker have in order for me to be able to hear back the recording of each of those bands with some hope that it sounded the same as when recorded and played back on the sound engineer's loudspeakers? Once the reference SPL was known and was set, if my speakers are anything other than flat, then I would not hear what the sound engineer heard (assuming that our hearing was similar).
Aren't these "voicings" simply sound colorations (i.e., distortions)? A sound coloration that you find desirable may be heard as deleterious by someone else. It doesn't really seem to be following a high fidelity course to have "voicings", which appear to be a marketing option.
Each recording has a "sweet spot" where it sounds best when reproduced at a particular sound pressure level (SPL). That's due to our auditory system's sensitivity function, as exemplified by the Fletcher–Munson curves and their more modern counterparts.
The sound engineer creates a mix using monitor loudspeakers that, as a general rule, aim for a flat on-axis frequency response. That mix is created at a certain SPL. That value of SPL is what bakes in the behavior of our auditory sensitivity function. So, don't expect to listen to loud music quietly, or vice versa, as the frequency balance as heard by the sound engineer is not being replicated as accurately as it could be. Hence the availability of loudness contour controls on amplifiers, as well as the bass and treble controls where we can adjust things to our liking. This is made easier if the loudspeaker has an un-voiced, flat response.
Of course, all of the above is somewhat of a simplification. Loudspeaker directivity and room acoustics will also affect our perception of reproduced sound in a listening environment.
Last edited:
(a) middle-ear transfer function (b) free field to ear drum transfer function (c) diffuse field to ear drum transfer function (d) outer and middle-ear transfer functions for the three listening conditions: free field, diffuse field, and headphone.
https://www.researchgate.net/figure...um-transfer-function-c-diffuse_fig3_269099439
However, I think all this is adjusted for during mastering and therefore we only need to reproduce things as they are.