Smooth (Flat) vs. Accurate (Hi-Fidelity)

Status
Not open for further replies.
there is no measure of accurate available

The only such measure could be "how it sounded to the mixing engineer"...

The sound waves emanating from a singer in a natural space are completely independent of the mixing engineer. Now the mixing engineer can add his own layer of deviation to that first output and I'll agree with you that the further we move down that path the less 'accuracy' becomes a coherent concept. Perfect reproduction of a deadmau5 track might be incoherent by definition.
Getting back to the singer, and assuming non-destructive distribution, if the recording engineer strives for absolute natural minimalism and doesn't try to modify the original acoustic event by diddling 'enhancement' knobs then the studio reproduction system and home reproduction system are just independent windows into the same mic feed. Which is more accurate returns to which sounds more like the singer in the space.
 
Thank you bigE and and ssZero for your support. Your comments are exactly what I had in mind to respond with to graaf upon my return to this forum after an extended absence.
The test I described and demonstrated is purely an objective one. Signal out vs signal in. Of course this can't be the only test of a speaker's ability to satisfy a listener. All the other measureable attributes researched by Toole and Olive do a very good job of relating objective measurements to subjective performance.
 
I am not sue what "a single test" means, but Drs. Toole and Olive would contend that sufficient tests of "good" versus "bad" do exist and I would tend to agree, although I might not agree with the exact tests and their interpretation.

I mean that you cannot test one aspect of a speaker and declare that it's good or bad based upon that result.

In this case, only testing how accurately a speaker reproduces a tone at a specific frequency does not tell us how the speaker will perform overall. For instance, it would not tell us anything about the performance of the crossover, like whether the crossover is optimized for the given drivers/enclosure, or how well the speaker holds up at high volume, overall frequency response, etc. Each of those would be their own separate test that could be performed and later used to determine overall speaker quality.
 
I mean that you cannot test one aspect of a speaker and declare that it's good or bad based upon that result.

In this case, only testing how accurately a speaker reproduces a tone at a specific frequency does not tell us how the speaker will perform overall.

But you would agree that multiple different tests of these various aspects CAN determine 'good" from "bad"? That's not obvious from either post.
 
1. Flat on-axis fr
2. low coloration (clean waterfall)
3. extended lf-bandwidth - ability to faithfully produce the lowest octave(s)
4. Dynamic capability - the ability to reproduce large peaks within bandwidth without highly rising distortion or physically breaking.
5. Good polar behavior - off axis fr is smooth with only a falling response for high frequencies

Thats a list i can come up with from the top of my head.
Regards,
 
the house curve totally transformed my system. the gradual roll off from 1khz to -6db at 20khz really did tame the brightness without any indication the sound is dull.
thanks a lot!!!
Hi Philosophil,

Wrt to frequency response, one would think the most accurate would be ruler flat. However, as Olive found out in his, “The Subjective and Objective Evaluation of Room Correction Products” Audio Musings by Sean Olive: The Subjective and Objective Evaluation of Room Correction Products , “flat in-room response is not the preferred target”. So what then is the preferred frequency response target?
 
the house curve totally transformed my system. the gradual roll off from 1khz to -6db at 20khz really did tame the brightness without any indication the sound is dull.
thanks a lot!!!

Glad you found the info as useful as I did.

Toole/Harman recently put out another paper that appears to be open access from AES: Journal of the AES 2015 July/August - Volume 63 Number 7/8

Check out Figure 14 for more on "Subjectively-preferred steady-state room curves" Note the trained listeners preference is more or less the same house curve as described in the other references. At the very least, a good place to start 🙂 Cheers!
 
Glad you found the info as useful as I did.

Toole/Harman recently put out another paper that appears to be open access from AES: Journal of the AES 2015 July/August - Volume 63 Number 7/8

Check out Figure 14 for more on "Subjectively-preferred steady-state room curves" Note the trained listeners preference is more or less the same house curve as described in the other references. At the very least, a good place to start 🙂 Cheers!

im totally impressed by the huge step up it brings.
musicality is really increased. I now find my self not wanting to go to bed and keep listening to music.
it also, with the ''house curve'' applied, make the speaker system sound very close tonally to my hd 650 headphones.
I guess that flat speakers may be really too bright!

highly recommended.

one question, can you tell em if my curve seems right?
im flat to 1khz then a straight line roll off:
-2db at 2khz
-4db at 7khz
-5db at 10khz
-6db at 20khz
 
Last edited:
Glad you found the info as useful as I did.

Toole/Harman recently put out another paper that appears to be open access from AES: Journal of the AES 2015 July/August - Volume 63 Number 7/8

Check out Figure 14 for more on "Subjectively-preferred steady-state room curves" Note the trained listeners preference is more or less the same house curve as described in the other references. At the very least, a good place to start 🙂 Cheers!

Thanks for posting that Mitch, it's damn near the curve I ended up with by experimenting a lot. I'd say getting the timing right helps too!
 
If you are into computer audio, then JRiver's 64 bit audio engine is as good as it gets in my opinion. Fab Filter looks good too.

In the digital domain, there are many good free and pay for eq plug-ins/packages. Ranging from simple eq to sophisticated psychoacoustic filtering in both the frequency and time domain. My experience is similar to wesayso's where getting the timing right helps too.

youknowyou, don't know if you use REW or not, (great piece of software by JohnM), but it has a facility to draw a target curve where you can look at the values you posted to see if they line up.

In addition, REW now has a new frequency dependent windowing and psychoacoustic smoothing option that more closely matches what we hear when measuring in room responses at the listening position. I loaded the target we are discussing in REW and then used a digital loopback measurement to have REW's sweep output go through JRiver's eq (in my case Convolution engine) and out through the DAC to amps and speakers. This was so I can validate that the speaker measurement at the listening position matches the target. As you can see in this post it matches almost perfect: Feature Request: Frequency Dependent Windowing - Page 10 - Home Theater Forum and Systems - HomeTheaterShack.com
 
The sound waves emanating from a singer in a natural space are completely independent of the mixing engineer. Now the mixing engineer can add his own layer of deviation to that first output and I'll agree with you that the further we move down that path the less 'accuracy' becomes a coherent concept. Perfect reproduction of a deadmau5 track might be incoherent by definition.
Getting back to the singer, and assuming non-destructive distribution, if the recording engineer strives for absolute natural minimalism and doesn't try to modify the original acoustic event by diddling 'enhancement' knobs then the studio reproduction system and home reproduction system are just independent windows into the same mic feed. Which is more accurate returns to which sounds more like the singer in the space.

As soon as you use a mic your changing the sound. Mic selection, placement, angle all change the sound.
 
If you are into computer audio, then JRiver's 64 bit audio engine is as good as it gets in my opinion. Fab Filter looks good too.

In the digital domain, there are many good free and pay for eq plug-ins/packages. Ranging from simple eq to sophisticated psychoacoustic filtering in both the frequency and time domain. My experience is similar to wesayso's where getting the timing right helps too.
fab filter offer three phase options...

I'm not sure if I understand what you mean about timing though..





I highly encourage people to try the B & K curve, the Toole/Olive curve or flat up until 1khz then a straight line arriving at -6db at 20khz.

the difference in the presentation is staggering for the better for all of my systems. my continuum from jeff bagby and amphion one18 are both much more musical to listen to.
I'm at a point where im playing music to see how they will sound with the curve added, just like if I just purchased a new speaker.

musicality and emotional response is definitely increased, finding also that Im able to play my music more loud with all the benefits.

I highly encourage people to try and add a similar curve, im sure many system could benefit!
 
Last edited:
These findings look at lot like the old B&K falling curve for best in-room response. Everything old is new again.

Something I'm surprised no one mentions, is HRTF or head shading.
When we measure with a microphone, we measure at a single point. But when we listen to a speaker, we hear it with the on AND off axis ears. There is a lot of HF shading happening at the off axis ear because the head blocks it. And what does that combined on and off axis ear response look like? Surprise, surprise, a response that falls above ~1.25K dropping to -6dB at 20Khz. Sound familiar?

Of course heads are different, so that's just an average. But it seems to me that the falling response curve - as measured at a microphone - is very similar to the falling response that 2 ears would have for a speaker located 30 degrees off axis.
 
Depends what you are trying to do…

dave

Example implementing a 3-way build design using dsp and multi-amping using REW for measurement data.

Do i concentrate on each individual driver @1m then adjust DSP accordingly

Or position the mic at listening posistion and then implement DSP adjustmunts/cross-overs etc

Forgive me but I struggle to understand the requirement to try and acheive a flat response @1m
 
Status
Not open for further replies.