does anyone enjoy flat sound from audio speakers?

[boxerfan88]

seems like everyone doesn’t like flat sound

flat anechoic or flat in-room?

 

[StigErik]

It is not a target to be implemented, it is a preference curve indicating what people is likely to prefer based on a study where 11 people participated.

Well... I like to implement my preference....

 

[dominikz]

Looking at these 5 pages I casually estimate the average period of this thread is approximately 10 posts before the discussion restarts at the beginning and we go through the cycle again...

 

[Exaybachay]

What do you mean by constant? Do you have a measurement from your main listening position?

"Constant" means "constant with minor deviations and no obvious and intentional upwards or downwards sloping, no boosted bass or such". I think "flat" probably means the same, but to me it is a weird description of a constant graph.

 

[StigErik]

Flat enough for me. Measured in the main listening position @3meters. 1/6 oct smoothing.

 

[ahofer]

Flat enough for me. Measured in the main listening position @3meters. 1/6 oct smoothing.

View attachment 353493
View attachment 353494

Looks excellent. I am not as experienced as others with the lower graph representation, but it looks like there is both a trough and a resonance or mode at 120Hz (and maybe another just over 200Hz). Is that EQ correction fighting a room mode?

 

[StigErik]

Looks excellent. I am not as experienced as others with the lower graph representation, but it looks like there is both a trough and a resonance or mode at 120Hz (and maybe another just over 200Hz). Is that EQ correction fighting a room mode?

To the last question - yes.

 

[mglobe]

There's a fundamental issue here, which is addressed in the "Flat does not equal flat" video that @staticV3 linked to on page 1, but which has not been clearly stated here in the thread:

That issue is that there are two different kinds of "flat" response from a speaker: anechoic flat response, and in-room flat response.

A speaker that is designed for maximum fidelity will have flat anechoic response: that is, in a room specially treated to have virtually no reflections - or these days using special computerized test equipment that can simulate such a room - the speaker's frequency response will be very close to flat from the lowest bass frequency it can manage, up to 20kHz.

The important thing to understand is that in any real-world room, an anechoically flat speaker is going to produce a somewhat downward sloping frequency response: higher in the bass, and then sloping down somewhat through the midrange and the treble. This is the in-room response.

So no, pretty much no one likes flat in-room response because it sounds too bright. And it sounds too bright because, as noted by others already, higher frequencies naturally get absorbed by the walls, furniture and other stuff in your room than do lower frequencies. So what we hear is always "tilted down" relative to what comes out of the speaker.

But many people do like flat anechoic response. In fact, the existing research, which you can find cited here in many threads, suggests that in blind testing most people seem to prefer speakers that have fairly flat anechoic response.

Finally, all of the above is also impacted by a speaker's dispersion pattern. Some speakers produce most of their volume (energy) in a more narrow area radiating out at angles to the left and right of the speaker drivers, while others produce most of their energy in a wider area - a broader angle. This can impact the perceived soundstage, but also the perceived frequency response at a given listening position. My understanding of the research is that a majority of people prefer wider dispersion, although this preference is not as strong as the preference for flat/neutral frequency response, and while linear anechoic frequency response is considered a sign of high fidelity and good design, wide vs narrow dispersion is generally considered more of a preference thing than a better vs worse thing (as long as the dispersion is relatively constant and even throughout the frequency range from about 200-300Hz upwards).

/thread

 

[terryforsythe]

Is there a good link to “the Harman curve”, when I google it all the results are about headphones.

I downloaded the original curve from Harman about 2 years ago. I don't know if it still is up. Anyway, I put it in Google Docs and made it accessible. I also put in a version to use with Dirac Live, and a version with a slightly lower bass boost. Here are the links:

1. Harman Curve from Harman: https://docs.google.com/document/d/1ec8hDo05xArvyLZBU-anEaeC9NKc5uNCUSn6CGGKECQ/edit?usp=sharing

2. Harman Curve formatted for Diract Live: https://docs.google.com/document/d/1gaV41ET31kCk48tjv7RAliaE1CDo74Ta15KgeTrWyic/edit?usp=sharing

3. Harman Curve formatted for Dirac Live, with 2.5 dB less bass boost: https://docs.google.com/document/d/1_dkO7PTNOG3ovfEU9Bdian0bsZ1MgFA9riGYwGm5DzA/edit?usp=sharing

 

[terryforsythe]

Ok everyone speaks about the Harman curve! What is that and how do do get it?

See my post above.

 

[terryforsythe]

It is not a target to be implemented, it is a preference curve indicating what people is likely to prefer based on a study where 11 people participated.

That may have been the initial study. During an interview, Floyd Toole's assistant explained that the study was repeated using trained listeners and untrained listeners. Plots were presented showing the preferences of different listening groups. Although the results varied a bit, the general trend followed the Harman Curve if I remember correctly.

Anyway, the Harman Curve may sound good on some tracks, but not on others. There is no standard for mixing that is universally followed by people doing the mixing - sometimes mixes are lacking in the high frequencies, some mixes are lacking in the bass, some mixes are too bright, etc. It comes down to how the music was mixed and the listener's preferences. That is why I have more than one preset I use on my miniDSP.

 

[sigbergaudio]

Well... I like to implement my preference....

Only that's not really how it works.

Flat enough for me. Measured in the main listening position @3meters. 1/6 oct smoothing.

View attachment 353493

This is not flat, it's smooth.

Smooth = no major deviations. Flat = Equal energy across the the frequency spectrum. You have the former, not the latter.

 

[sigbergaudio]

"Constant" means "constant with minor deviations and no obvious and intentional upwards or downwards sloping, no boosted bass or such". I think "flat" probably means the same, but to me it is a weird description of a constant graph.

So when you play at say 75dB, both 50hz and 10khz is at roughly 75dB when measured at the listening position?

 

[sigbergaudio]

That may have been the initial study. During an interview, Floyd Toole's assistant explained that the study was repeated using trained listeners and untrained listeners. Plots were presented showing the preferences of different listening groups. Although the results varied a bit, the general trend followed the Harman Curve if I remember correctly.

Anyway, the Harman Curve may sound good on some tracks, but not on others. There is no standard for mixing that is universally followed by people doing the mixing - sometimes mixes are lacking in the high frequencies, some mixes are lacking in the bass, some mixes are too bright, etc. It comes down to how the music was mixed and the listener's preferences. That is why I have more than one preset I use on my miniDSP.

While I agree there are certainly some differences, mainly between periods (50s music is pretty different from something mixed in 2010) - I would argue that a well balanced system will work well for most tracks without the need for changing the balance from track to track.

 

[Theta]

Great explanation, thank you. Now, with all this knowledge and all these measurements, why can't engineers design a decent speaker for a fair price, why do some speakers you can pick up on the used market for a couple of hundred bucks sound as good or better then new speakers in the 10 to 15K? Measurements are great, subjective opinions are unreliable at best Results are poor in both cases.

 

[terryforsythe]

While I agree there are certainly some differences, mainly between periods (50s music is pretty different from something mixed in 2010) - I would argue that a well balanced system will work well for most tracks without the need for changing the balance from track to track.

I wouldn't call it a "need", more of a preference.

There is no audio system that is going to play every track for the optimum preference of all listeners without equalization. Given the wide range of mixing styles used when mixing music, the wide range of mixing equipment used, and differences in equipment setup, it is impossible.

For example, consider a person (mixer) who mixes a track using monitors with a flat frequency response as measured at 1 m, and where those monitors are 1 m from his/her listening position while they are mixing the track. Also assume that the final mix is optimized for the mixer's tonal preferences.

Now, assume a listener has those same tonal preferences as the mixer, and also is using speakers that measure flat at 1 m, but in this case the speakers are positioned 3 - 4 m from the listening position. The in-room response at the listening position will be different than the response at the mixer's listening position, and the track will be lacking in the high frequencies. For that scenario, having a preset where the system is tuned for a flat response at the listening position may improve the sound from the perspective of the listener.

Take the same example, but this time assume that the mixer has the monitors positioned 3 - 4 m from the listening position while tuning the final mix. The in-room response at that listening position will be different than at 1 m, it may be more in line with the Harman Curve above the bass frequencies, depending on the room characteristics. If the listener tunes their system for a flat frequency response at his/her listening position, the track will sound too bright.

 

[sigbergaudio]

For example, consider a person (mixer) who mixes a track using monitors with a flat frequency response as measured at 1 m, and where those monitors are 1 m from his/her listening position while they are mixing the track. Also assume that the final mix is optimized for the mixer's tonal preferences.

Now, assume a listener has those same tonal preferences as the mixer, and also is using speakers that measure flat at 1 m, but in this case the speakers are positioned 3 - 4 m from the listening position. The in-room response at the listening position will be different than the response at the mixer's listening position, and the track will be lacking in the high frequencies. For that scenario, having a preset where the system is tuned for a flat response at the listening position may improve the sound from the perspective of the listener.

Take the same example, but this time assume that the mixer has the monitors positioned 3 - 4 m from the listening position. The in-room response at that listening position will be different, it may be more in line with the Harman Curve above the bass frequencies, depending on the room characteristics. If the listener tunes their system for a flat frequency response at his/her listening position, the track will sound too bright.

While I again agree that yes there are certainly differences between tracks, I think you are selling the engineer a bit short here.

Any mastering engineer worth their salt understands that this will happen, and will compensate accordingly when doing the mix, and if needed also check how it translates by testing the mix on different systems.

 

[SuicideSquid]

Only that's not really how it works.



This is not flat, it's smooth.

Smooth = no major deviations. Flat = Equal energy across the the frequency spectrum. You have the former, not the latter.

Looks like a flat anechoic response that gives a slightly tilted in-room response to me. Or, "flat".

 

[sigbergaudio]

Looks like a flat anechoic response that gives a slightly tilted in-room response to me. Or, "flat".

This is a pretty confusing concept to most, so I think we need to be explicit. "Flat" is not helpful. Tilted is not flat, it's tilted.

The curve presented is not flat in any form or shape. Will the speakers in question measure flat anechoically? We don't know.

 

[terryforsythe]

While I again agree that yes there are certainly differences between tracks, I think you are selling the engineer a bit short here.

Any mastering engineer worth their salt understands that this will happen, and will compensate accordingly when doing the mix, and if needed also check how it translates by testing the mix on different systems.

Unfortunately, I sometimes come across music I like, but is not recorded by a mastering engineer worth their salt.