Share your in-room measurements?

[Mike Lima]

View attachment 138288

View attachment 138289

Your thoughts? Please give me constructive ideas on what I can do. I am very much a newbie. Thank you.

Bueller?

 

[boba]

I was lucky enough to find a cheap used set of Dali Spektor 1 speakers along with a free BIC Venturi V1020 subwoofer. I'm using these for a computer desk in the corner of a large room with no treatment. The speakers are raised off the desk with various hard and soft foam blocks, which aligns the tweeters just below ear level. The speakers are 52" (130 cm) apart and the listening position is 36" (90 cm) from each one. The subwoofer is on the floor at the same distance. The speakers are toed in slightly by 20 degrees, but due to the close listening position they're still 25 degrees off axis. This reduces the "bright" sound considerably (as mentioned in the review here).

I used a UMIK-1 microphone on loan, pointing it at the phantom center to do measurements. EQ was calculated using the average of 7 responses for each speaker: listening position and 4-inch offsets (forward, back, left, right, up, down). For the house curve, I used a rough approximation of the "subjectively-preferred" curve described in this paper (p. 528).

Measurement before EQ:

Measurement after EQ:

Waterfall before EQ:

Waterfall after EQ:

Not perfect and I'm still learning, but it works for me.

 

[Thomas_A]

I just make a note that several of you seem to room-correct above Shroeder frequency. That will mess-up the first-arriived on-axis frequency response of the speaker.

 

[czt]

ARC3 gives me better results than MathAudio (or REW with Eq. APO, DRC-FIR with DRCDesigner and Convolver is unusable because of volume fluctuation):

 

[Chromatischism]

ARC3 gives me better results than MathAudio (or REW with Eq. APO, DRC-FIR with DRCDesigner and foobar2000 Convolver is unusable because of volume fluctuation):

View attachment 148900
View attachment 148901

I want to use that on the PC but the company tells me the correction can only be used in a DAW.

 

[czt]

As you can see it's used through jBridge (with "Sluggish GUI hack" setting) in foobar2000 with VST 2.4 adapter as the VST is unfortunately only 64 bit, and ASIO4ALL for the Analysis because of UMIK-1. It was a hassle to use three other tools to make it work, I almost gave up on the first attempt.

 

[holbob]

I just make a note that several of you seem to room-correct above Shroeder frequency. That will mess-up the first-arriived on-axis frequency response of the speaker.

Is it better to correct above schroeder if you listen off-axis? I have Elac dbr62 which are designed to be listened to off-axis (I have them straight forward)

 

[Thomas_A]

Is it better to correct above schroeder if you listen off-axis? I have Elac dbr62 which are designed to be listened to off-axis (I have them straight forward)

Generally no. Correcting above Shroeder should be done only if you correct for speaker response. Not speaker-room response.

 

[Chromatischism]

Is it better to correct above schroeder if you listen off-axis? I have Elac dbr62 which are designed to be listened to off-axis (I have them straight forward)

A lot of it has to do with not trusting these systems to make the right decisions in terms of what to EQ and what not to EQ when given room/power response data, including all the small wiggles that our brains average out.

Also, the more things in your room causing odd reflections (blooming off a coffee table, for example), the more the system will reduce output from the speaker in the affected frequency ranges. This also puts a dip in the direct sound you are getting from the speaker, so it's not ideal.

 

[rdenney]

Makes me think a flat REW plot has become the target, rather than balanced and appropriate sound.

As Toole suggests, one should adjust using equalization above the Schroeder frequency only based on anechoic data.

My own, without and with EQ, with adjustments only up to 300 Hz:

There are wiggles I can’t hear and my space minimizes the early reflections that would fit the Harman model. Speakers are Revel F12s.

Rick “EQ games” Denney

 

[boba]

I just make a note that several of you seem to room-correct above Shroeder frequency. That will mess-up the first-arriived on-axis frequency response of the speaker.

Generally no. Correcting above Shroeder should be done only if you correct for speaker response. Not speaker-room response.

This is interesting. Do you (or anyone else reading) have a link to a paper or article from a good source that goes into detail about this? I did a quick search and just found a lot of passionate arguments for and against correction above Schroeder frequency. I don't mind redoing measurements if it means better results.

 

[Balle Clorin]

I posted in post #77, But I have changed speaker distance and listening distance and did a trick with a door

Speakers Revel F36, no sub

Old setup

New setup. Speakers 150cm from frontwall, listening at 430cm from front wall. Room 600cm long 380cm wide(408-full wall bookcase) 313cm high
Speakers 80-60cm from sidewall
Yellow no DSP, DSP full range=Blue one single point measurement, orange=one measurement with moving mic method and DSP full range

Look quite flat compared with Harman slope ,but does not sound too bright for my 60+ears, But I do lift the bass 2-3 db more below 200 than in this plot. I found that full range DSP sometimes work better than only below 1k, but the jury is still out,,,

 

[Chromatischism]

Makes me think a flat REW plot has become the target, rather than balanced and appropriate sound.

As Toole suggests, one should adjust using equalization above the Schroeder frequency only based on anechoic data.

My own, without and with EQ, with adjustments only up to 300 Hz:

There are wiggles I can’t hear and my space minimizes the early reflections that would fit the Harman model. Speakers are Revel F12s.

Rick “EQ games” Denney

I know you meant smooth and not flat, but to clarify for everyone else, flat in-room, from your listening seat a few meters from the speakers, is not the target. At least not above 2-3 kHz. To get a flat measurement all the way through the treble there often means the speaker itself is boosted above neutral. Note that is not the case if you are near field, which should measure much closer to flat without any EQ.

You have a good starting point - now with a rise to the bass starting at 100-120 Hz or so and increasing about 6 dB to 20 Hz, you will have more balanced sound. The reason I don't extend the bass boost beyond 120 or so is to maintain clarity - you don't want those instruments and vocals to start sounding boomy or congested.

 

[boba]

Here's the curve I mentioned earlier. Is there anything wrong with using the "trained listeners" line as a house curve in REW, but with a slight boost in bass to match my personal preference? That's the purple line in my previous post. I also thought that you should never EQ to a completely horizontal line in a typical reflective room (i.e. not an anechoic chamber), but I see some people doing that in this thread. @Chromatischism has mentioned that as well so maybe I'm not alone.

 

[Chromatischism]

Yup, the closer you get to the speakers the more your measurements should resemble the listening window of the speaker.

 

[rdenney]

I know you meant smooth and not flat, but to clarify for everyone else, flat in-room, from your listening seat a few meters from the speakers, is not the target. At least not above 2-3 kHz. To get a flat measurement all the way through the treble there often means the speaker itself is boosted above neutral. Note that is not the case if you are near field, which should measure much closer to flat without any EQ.

You have a good starting point - now with a rise to the bass starting at 100-120 Hz or so and increasing about 6 dB to 20 Hz, you will have more balanced sound. The reason I don't extend the bass boost beyond 120 or so is to maintain clarity - you don't want those instruments and vocals to start sounding boomy or congested.

With respect, I disagree. The downward spectral tilt is a byproduct of a flat speaker (and I do mean flat) in a typical room, where the bass is omnipresent but early reflections don’t see as much of the treble because of treble directivity. A room with relatively little early reflection (like mine, because of its shape) will provide more treble than listening positions more dominated by reflected sound.

Toole has said this a number of times. The downward tilt is a room effect, not an objective. It indicates an anechoically flat speaker in a typical room.

Whether people like more or less bass is a matter of preference. A preference model may or may not apply to an individual. As a tuba player, I have a specific notion of what I want bass to sound like.

Rick “who has checked this knowledge several times” Denney

 

[Chromatischism]

With respect, I disagree. The downward spectral tilt is a byproduct of a flat speaker (and I do mean flat) in a typical room, where the bass is omnipresent but early reflections don’t see as much of the treble because of treble directivity. A room with relatively little early reflection (like mine, because of its shape) will provide more treble than listening positions more dominated by reflected sound.

Toole has said this a number of times. The downward tilt is a room effect, not an objective. It indicates an anechoically flat speaker in a typical room.

Whether people like more or less bass is a matter of preference. A preference model may or may not apply to an individual. As a tuba player, I have a specific notion of what I want bass to sound like.

Rick “who has checked this knowledge several times” Denney

I'm not sure where the disagreement would be because that is also my understanding.

 

[GDK]

Stupid question from me: why do we measure L and R separately, rather than LR together? When I measure them both together (post-EQ), I get very different results than an average of the two. In particular, there are often large cancellations evident that I can only get rid of by smoothing the response even more.

 

[abdo123]

Several subwoofers later, I'm finally satisfied with my bass response. (1/12 smoothing).

There is nothing i can do at the moment about the 60 to 80 range because that's a floor/ceiling bounce and i would need to lift some subwoofers to really deal with it. I will probably will, just not right now.

Barely any aggressive EQ, Distortion was maintained below 1%

 

[Balle Clorin]

Stupid question from me: why do we measure L and R separately, rather than LR together? When I measure them both together (post-EQ), I get very different results than an average of the two. In particular, there are often large cancellations evident that I can only get rid of by smoothing the response even more.

The waves from each speaker create interference giving cancellations and the opposite. That is why especially the treble can show crazy deviations . Move the mic or one speaker 2cm and the plot can change dramatically. If you run random pink noise and plot RTA instead of Spectrum you avoid this.