Post Your Frequency Response Curve!

[Tim Link]

Below you can see two measurements I took last night. This was taken at the listening position, which is 10 feet back from the center speaker and about 12 feet from the woofer stacks. The room is 12 feet wide by 20 feet long and about 7.5 feet tall. The green curve on top doesn't sound as good as the red one below. I had noticed my system was sounding really good the night before. When I took the measurement last night I felt it wasn't sounding as good, and the curve I saw was not what I expected. It's extremely flat from about 30 to 200, and then jumps down and is flat again from about 600 to 6K. I recalled it having a more continuous downward slope. So I restarted all the digital equipment and re-measured. The red curve is the measured result after restart and it sounds good and looks right. I should note I took these measurements repeatedly before and after the reset, and they stayed consistent with slight changes in the microphone position or moving things around in the room. So I know it's the sound coming from the speakers that's changing. This confirmed something I suspected might have been happening, and led me to all kinds of confused notions about different amps, digital sources, etc. sounding different. Now I don't know. It might just be that restarting the system fixes something that went off. Over the years I've noticed very obvious distortions creeping in from time to time that require restarting or reconnecting things to get it straightened out. It made me wonder if perhaps more subtle things might also be happening and messing with my mind. I think this verifies it. I can't for the life of me understand why this frequency response change would occur but when software is involved the possibilities are limitless.
I use cheap digital equipment. I'm not sure if all cheap digital equipment has problems like this but I'd be willing to pay more for stability.

 

[MarkusBarkus]

@Tim Link I think restarting equipment once-and-a-while does have an effect. I'm not tracking this possible effect in my system, but anecdotally, I think there is something to it. IIRC somewhere in the Taiko Extreme thread, Emile and others stating a restart helps SQ. I have heard this re: routers and modems as well as audio gear. Certainly, Netflix needs it once and a bit. Somehow, "things" seem to like a deep-breath/re-set.

BTW: your room does NOT look 7.5 feet tall. As an owner of a sub-eight footer, I always wonder about that. Cheers, fella...

 

[Tim Link]

@MarkusBarkus You have a keen eye! My room is actually very close to 7' 8", about 7.7 feet tall. I was worried when I first moved here that my speakers weren't going to fit. That would have been a mess.
I'll take some comfort in knowing that I'm not facing digital gremlins alone. I'm now turning off my computer after a listening session instead of putting it to sleep.

 

[Tim Link]

I did a lot of listening last night, comparing different contours for the bass. Like others here, I only EQ the bass region. The curves are highly smoothed to show the trend better. The orange is what I started with and the blue is what I ended with. The reduced bass sounds tigheter and brings more life to the midrange, which was getting swamped. I also moved the crossover for the woofers up from 200 to 300Hz. I'd go higher but above that the response drops off a cliff. Ultimately I took the CinemaPanels down. This room already has a pretty dry sound with these Revels set up like they are and it needs all the liviness it can get. The CinemaPanels I tried are too absorptive for here. I'll try the SoundPlanks, which are not as wide and have a larger perceentage of their surface reflective. I should have put a hard floor in here with a rug in the middle instead of wall to wall carpeting.

 

[Tim Link]

Very crude, using iPhone 13, just placing the phone in front of my face and leaning back so it is “about” where my head is. I think I am using the same pink noise track MikeL identified. I have some others off Stereophile test CDs.

A weighting:
View attachment 104735

And Flat weighting:
View attachment 104736

Whats that mean? (Besides a broad dip ~100 hz)

I see a nice slope that suggests your system is working well in your room. It looks like there's a bit of weak response around 110hz at your measuring position. Hard to say if it means anything important audibly. If I hear what I think is a strangely loud or weak note while listening to music I'll check the spectrum anlayzer and see what frequency that note is. If I can, I'll EQ it until it sounds right. I've done this by ear method and then measured my results later and they were pretty much spot-on in terms of narrow band peaks and dips. I find broad band adjustments harder to do by ear.

 

[Tim Link]

I'm experimenting with a flatter response and a higher crossover setting to the horn woofers. Not the Harman curve here. FLAT! Well, there is some roll off above 5K. I'm still not doing any EQ on the Revels so I'm not sure why their response looks flatter than before. It's bright sounding for sure. This is also very dynamic. Wow. There's a lot to like about this sound. It's really bright but I haven't gotten tired of it. The little Revels seem to really like being relieved of duty at 600Hz. The continuity of the bass on the woofers extending up higher does something very good to the bass clarity too. Some people think an 18" woofer isn't fast enough to do 600Hz well. My ears are telling me it doesn't work that way. They're a lot better at it than a little 6.5" driver if you're trying to deliver any kind of volume. They also hold some pattern control down into that range. It sounds really clean and fast.

 

[sigbergaudio]

I don't want to go too far off topic but I agree proper diffusion, or sound scattering is very helpful at first reflection points and throughout the room. Last night I took home four panels that are about 8" wide x 48" tall and two inch thick. These have a reflective strip down the center which scatters sound, creating a new line source of reflected sound at that point. The edges of the panels also scatter sound, which is why dividing a single larger panel into some thinner strips and spacing the strips apart on the walls can work some magic. What I heard last night and this morning with those panels temporarily propped up along the sidewalls was a smoother overall tone and some even more magical imaging than I was already getting with my special 3 speaker array.

To get back closer to the topic of this thread, I didn't see a quickly noticeable measurement difference from these panels in terms of overall frequency response, but I'm hearing the imaging improvements and some perceptual reduction of the room's characteristics.
I'll post some measurements next.
View attachment 105273

That looks like an incredibly narrow setup. Those left/right speakers should probably be at least 3 feet further to the side each?

 

[Tim Link]

That looks like an incredibly narrow setup. Those left/right speakers should probably be at least 3 feet further to the side each?

Ah you noticed! Yes, this is an unusual setup that I've been experimenting with. I'm using 3 speakers for 2 channel playback. The speakers are close together because they produce a crosstalk canceling effect while maintaining a rock solid center image that to my ears just sounds better than a phantom center image. The center speaker plays L+R signals mixed. The left plays L-R and the right plays R-L. This is similar to Hafler's old method for simulated surround sound. What I discovered is that if you place these three speakers close together in a row they produce very good imaging. I initially had the speakers all jammed up right next to each other, trying to get the drivers distances close to ear distance apart because that made sense to me. I was surprised to discover that the strong stereo separation and sound field width was robust against moving the speakers further apart, which is fortuitous because that way the speaker baffles don't influence each other so much. I did some wave simulations to discover that the cancelation effects can work over a variety of speaker spacings so long as the listening position is an appropriate distance back. So far I'm really loving this and wonder if I'll ever go back to just using two speakers for stereo. The sound field extends well past the corners of the room, at times reaching almost 180 degrees wide.
To mix the channels I'm using a computer. This could be done with an analog circuit and I might try to implement that because it'd get the computer out of the signal chain. Also it can be done with just a two channel system. The two side speakers can be wired in parallel, with the right speaker wired out of phase. So the left channel gets mixed to L-R and the right channel gets mixed to L+R. Send the right channel to the center speaker and the left channel to the side speakers. Just make sure your amp can handle the reduced impedance from wiring the two speakers in parallel. It ends up working out pretty well for the amp's sake because the L-R signal tends to be weaker than the L+R, so I think the total workload on each channel ends up being very similar.

 

[sigbergaudio]

@Tim Link Huh. Interesting approach.

 

[Tim Link]

@Tim Link Huh. Interesting approach.

I recommend you try it! I'm curious to know if other audiophiles find it as compelling as I do. So far I've managed to get a few people to listen to it. My non-audiophile friend thought it sounded good. I guess that's better than bad! My boss thought it was intriguing but he seemed a little uncomfortable with the new presentation. He noted that it definitely worked, and could be quite useful in small spaces since all three speakers can be up close to each other. I have a setup on my office desktop, which is the one he listened to. I listen to it very close, like 2 feet back from the center speaker. He preferred to get back from it further to lessen the effect. If you get off axis the center stays put but the sound stage narrows and blurs away into a non descript ambient effect. It still sounds like stereo but you can't really place where anything is off to the sides of the sound stage. You can also play with the center to side level ratios to narrow and solidify the soundstage, or make it really wide and spooky. If you turn the center off you end up with a karaoke track so you can provide your own vocals!

 

[picears]

Idiosyncratically for me I seem to be very sensitive -- in the sense of finding it edgy or fatiguing -- to a material bump in the 2kHz to 4kHz range.

Have you ever tried in-ear frequency response measurements, and made (via DSP) corrections based on those? I'm thinking here of something like Professor Choueiri's software/device.

 

[Ron Resnick]

Hi,

I had my hearing tested a few years ago.

No, as I don't believe in DSPing a full-range analog signal.

 

[sbo6]

Unfortunately I cant use the App mentioned as I don't own any Apple products but I do have REW Charts.

Measurements shown are with DSP, 5 channels total (2 Main Towers, 3 Subs) and Audiolense Filters. Listening room is a 25ft x17ft x at least 10ft high and is an untreated Living Room (at least 4500cu/ft). Mic is 9ft away at seated listening position/seated listening height (Earthworks M30) and the rest of the measurement gear was a Lynx Hilo and Earthworks 1021 Mix Pre. Sweeps done at 24/192kHz. These were done using a Freq Sweep from 10-25kHz

This is closest thing I have showing zero smoothing of the FR with all channels combined. Ignore the lower chart showing Phase.

Closest thing I have showing the 10-350hz range without smoothing is this Waterfall

And here is an FR only plot. 1/12th smoothing, 60-90db range, individual L/R Channels as well as Both channels combined. Its easy to see the +6db of gain from the combined measurement here.

Maybe turn up those subs?

 

[picears]

I had my hearing tested a few years ago.

What Prof. Choueiri does is not a hearing test (the results of which depends upon the complete human hearing system) as your audiologist performed.

Instead, it's a response curve showing the effects of the outer ear and head shape on the sound that hits the eardrum. He uses that in the correction of the signal (along with cancelling interaural crosstalk.)

We all hear differently. This is something that so many people miss. It's not just that from the eardrum inwards we are different - we are different on the outside too.

I don't believe in DSPing a full-range analog signal.

In a digital system, the listener-specific adjustments are done in the digital domain, upstream from the DAC.

 

[Ron Resnick]

Very interesting! Thank you for explaining!

 

[cjf]

Maybe turn up those subs?

The Subs can certainly be cranked up well beyond what I currently have them set at but in my experience in doing that in the past (in the current room) is that it results in the THD numbers starting to rise very quickly (into the Double digits and even into the Triple digits in the Sub Freq ranges -20hz), especially when listening at elevated levels like I usually do.

My curves were generated based on a 75db Ref listening level. All channels were set to the same Output level prior to Room Correction filter creation. This was done by playing Random Pink Noise in the Lower Bass regions from each channel. In reality though, I listen quite a bit louder than that. So my expectation is that this curve probably flattens out quite a bit more once you start to get into the 90-105db range. The system does not sound bass deficient in any way to my ear. The Subs blend about as good as I could muster with the Main channels and are mostly unlocatable by ear into the SPL ranges previously mentioned.

But, if I really wanted to flatten the curve so to speak (without adding higher levels of THD in the Sub Bass region) I would just need to buy more Subs. The current room is fairly large and very open to every other room in the house. Its going to take at least 2-3 more large displacement Subs to get real rowdy in here. If I weren't moving in the future I would have already done that. Next house will be smaller so may not need the additional Subs then.

 

[Keith_W]

I'm experimenting with a flatter response and a higher crossover setting to the horn woofers. Not the Harman curve here. FLAT! Well, there is some roll off above 5K. I'm still not doing any EQ on the Revels so I'm not sure why their response looks flatter than before.View attachment 105406 It's bright sounding for sure. This is also very dynamic. Wow. There's a lot to like about this sound. It's really bright but I haven't gotten tired of it. The little Revels seem to really like being relieved of duty at 600Hz. The continuity of the bass on the woofers extending up higher does something very good to the bass clarity too. Some people think an 18" woofer isn't fast enough to do 600Hz well. My ears are telling me it doesn't work that way. They're a lot better at it than a little 6.5" driver if you're trying to deliver any kind of volume. They also hold some pattern control down into that range. It sounds really clean and fast.

Flat at the listening position is much too bright.

Speakers that measure flat under anechoic conditions will have a downward tilt when placed in a listening room. So if you make the FR flat at the listening position via DSP, what you are actually doing is applying an upwards tilt to the speaker. I have the ability to design my own target curve, and I much prefer a downwards tilting frequency response. In fact, I can look up the frequency response of any speaker and replicate that at home with a custom target curve. It won't sound the same as the other speaker, but the tonality will be very close!

 

[Tim Link]

Flat at the listening position is much too bright.

Speakers that measure flat under anechoic conditions will have a downward tilt when placed in a listening room. So if you make the FR flat at the listening position via DSP, what you are actually doing is applying an upwards tilt to the speaker. I have the ability to design my own target curve, and I much prefer a downwards tilting frequency response. In fact, I can look up the frequency response of any speaker and replicate that at home with a custom target curve. It won't sound the same as the other speaker, but the tonality will be very close!

Indeed, I generally prefer a downward tilt at the listening position too. I was just experimenting, and found it was interesting to hear a relatively flat combined response.

I understand that typical direct radiator box speakers that measure flat on axis should have a smooth downward tilting total room response. Omnis, dipoles, and corner horns might produce a flatter combined in-room response when their on-axis anechoic response is flat.

I'm finding that I seem to have preference for either highly directional speakers that maintain control down to a low frequency, like a corner horn, or a very small speaker that has a very wide dispersion through most of it's range. A 3/4" dome tweeter matched to about a 4" mid-woofer on a narrow baffle will usually gain my favor over a dome tweeter with a waveguide mated to a 6" or larger woofer on a wider baffle, even if the larger speaker sounds better in many ways. I think it's because the beam width difference is more dramatic between the bass and treble, resulting in a steeper downward slope at the listening position. At some point that slope gets too steep and it sounds a little dead to me.
I have no ownership experience with big ribbon line sources or planars. I suspect I might like those too.

 

[Final]

Pink noise graph recorded without weighting in my living room. I apologize for difference in app, but one gets the picture.

 

[Tim Link]

Pink noise graph recorded without weighting in my living room. I apologize for difference in app, but one gets the picture.
View attachment 106930

This is an interesting response curve, fairly flat with a slope starting at around 2.5K. What kind of speakers are these? Are you on-axis?