Not a bad result in a relatively small room right? I'll repeat the measurements when I've replaced my big poster with a damping panel. I'll have to redo the entire process anyway. But there will come a point at which I cannot add more damping to the room without being expelled by my girlfriend 
.
.
Last edited:
Well four, and a question:
First, measuring is waste of time
http://www.diyaudio.com/forums/full...t-current-speaker-measurement-technology.html http://www.diyaudio.com/forums/full...m-measurements-vs-no-measurements-ref-fh.html.Second, i won't visit you for a listening, chances you hang me out the forum too big
.Third, border to get envy exceeded several times.
Fourth, thanks the journey so far and sharing.
Have they reproduced measured squarewaves yet.
Those measurements look remarkably even through the bass. It's a little surprising because your line arrays should not alleviate the front-back mode in the room, right? The floor-ceiling mode could be eliminated, and maybe the left-ride mode. Is there any smoothing on the graph and did you apply EQ to the bass to remove that mode? It's helpful to see the graph of a scale of 50 db with 5 db increments.
You could try to hit the B&K curve, which gives a very pleasing and perceptually flat sound most of the time(http://www.bksv.com/doc/17-197.pdf). It is important to note that you shouldn't sacrifice a flat anechoic response to match some predetermined room curve. Your room curve will be governed by your loudspeaker's main axis response, loudspeaker directivity and room absorption. You should target flat anechoic and then let the steady state curve fall where it wants. Flat anechoic means a gated response. So maybe from 300 Hz onwards, you should try to hit a super flat response.
You could try to hit the B&K curve, which gives a very pleasing and perceptually flat sound most of the time(http://www.bksv.com/doc/17-197.pdf). It is important to note that you shouldn't sacrifice a flat anechoic response to match some predetermined room curve. Your room curve will be governed by your loudspeaker's main axis response, loudspeaker directivity and room absorption. You should target flat anechoic and then let the steady state curve fall where it wants. Flat anechoic means a gated response. So maybe from 300 Hz onwards, you should try to hit a super flat response.
Last edited:
Well four, and a question:
First, measuring is waste of time
Second, i won't visit you for a listening, chances you hang me out the forum too big
Third, border to get envy exceeded several times.
Fourth, thanks the journey so far and sharing.
Have they reproduced measured squarewaves yet.
No square waves recorded yet, the speakers are now playing as my home theatre for my kid and his friends watching Puss in Boots
.
I did listen to a lot of different curves so far. One of them being as flat as I could get the response. I equalise the room with speakers combined (no gating) as a system using DRC. So far that seems to work judging purely on a smiles per minute base
. It will be hard to get reflection free measurements using gating as at my listening distance the wall behind me is very close.
Last edited:
The increments of the graph is 5 dB actually, So posting it on a larger scale it would make it look more restless, not much else
. 
A few pages back I explained what I did in the bass, yes I do use EQ to get it like this. DRC is first used to even out the left and right response. After that there was a hole in my bass response:
Separate left and right responses didn't show that dip. Due to the use of a ground loop isolator I noticed a difference in the left and right channel. I don't know if you read that part but the ground loop isolator caused a rise in distortion at bass frequencies. But it also provided me with a clue how I might solve that. The left channel showed a rise in distortion around the 60-70 Hz dip area. The right channel didn't have that raised distortion at those frequencies. So I made a PEQ cut (after the convolved result) of -8 dB at 64 Hz to the left channel and an equal boost at 64 Hz on the right channel, both with Q=5. Furthermore a cut to the left channel of 4 dB at 71 Hz and the same as a boost to the right channel at 71 Hz (basically trading the bass in left and right).
As these frequencies are well below Schroeder you can't really sense where they are coming from anyway. At 32 Hz I do something similar but the other way around. Not to cure a dip but to relieve the right channel of some boost because the left channel has more corner gain and as such can produce those notes easier. The result was this:
It is remarkably stable over the entire listening area, acting like distributed bass I guess because it is being produced over the entire length of the array.
The bass is never boomy, easy to follow separate notes and sounds more like it would in a larger room.
To add more of a room curve I showed this earlier:
You'll notice it is flat from 110 Hz to 1 kHz and downwards slope above that. Based loosely on material I read from Bob Katz. B&K curve sounded more dark in comparison, not unpleasant though. This curve sounds sweet with most of the material I play. I'll revisit these curves after I add my damping panel behind the listening spot.
One note to add: I usually use the variable smoothing from REW's beta version. It smooths depending on frequency sort of like using a windowed result. There isn't much smoothing done to the bass using the new variable smoothing from REW, one could argue it is too smooth for higher frequencies.
The average graph posted one page back isn't smoothed.
Just documenting my journey here, it might help others in the future. I know I learned a lot on here from others kind enough to share.
Last edited:
Wesayso,
Very excellent result! It is remarkable how EQ the room dip at 60-70Hz did not increase the HD there - in fact, it looks even lower. How are you able to do that? Is it some DRC magic? Normally, when I boost a 7dB dip, there is almost a certain increase in HD at the same spot. This may be indicative of the fact that your 50 drivers are hardly moving at all near 70Hz, hence getting more boost out of them wasn't much more cone excursion into the HD zone of cone movement.
Very excellent result! It is remarkable how EQ the room dip at 60-70Hz did not increase the HD there - in fact, it looks even lower. How are you able to do that? Is it some DRC magic? Normally, when I boost a 7dB dip, there is almost a certain increase in HD at the same spot. This may be indicative of the fact that your 50 drivers are hardly moving at all near 70Hz, hence getting more boost out of them wasn't much more cone excursion into the HD zone of cone movement.
Well the truth is the distortion graph is still showing a lot of ambient noise so it would be even lower than what I get here. It changes per graph showing me it's not the minimum that could be obtained. I tried to run more sweeps in REW but REW screws up the timing adding up the results of those multiple sweeps. But think about it, a 8 dB boost, how much is one driver seeing from that amount? What I did here is trade a dip, that can't be boosted on the left for creating a peak on the right. So I'm not adding power to where it doesn't make sense. You always have to be careful to fill in dips. A simple boost does not work and sounds wrong. I was lucky enough to find a way to fill it in otherwise because I don't have a symmetrical setup.
I already had plans for multiple subs to even out the bass but apart from being fun for home theatre is isn't really needed for my music.
Even E40's music works very well. That music is really not my style but was touted as having a lot of low bass in the car forums. It seems to ride at ~40 Hz.
Same for the bass most people without a proper setup miss on the track: Gorillas - Clint Eastwood. Lorde, with Royals always gave me all kind of troubles in my Car, where things started to rattle, took me a while to cure that, again just above 40 Hz.
I bet real orchestral music is way more taxing on the low end but that isn't my thing. Not that I love the above mentioned tracks, I just know from experience they have an exaggerated amount of low bass from testing in my car.
Just played Fleetwood Mac's Albatross, with alternating left/right bass at ~62 and ~85 Hz, still sounds like it's coming from the sides, probably due to the harmonics above the fundamental tones. JRivers analyser is a great tool to see those tones especially since they are boosted in my case.
(bass in Albatros in analyser)
I already had plans for multiple subs to even out the bass but apart from being fun for home theatre is isn't really needed for my music.
Even E40's music works very well
. That music is really not my style but was touted as having a lot of low bass in the car forums. It seems to ride at ~40 Hz.Same for the bass most people without a proper setup miss on the track: Gorillas - Clint Eastwood. Lorde, with Royals always gave me all kind of troubles in my Car, where things started to rattle, took me a while to cure that, again just above 40 Hz.
I bet real orchestral music is way more taxing on the low end but that isn't my thing. Not that I love the above mentioned tracks, I just know from experience they have an exaggerated amount of low bass from testing in my car.
Just played Fleetwood Mac's Albatross, with alternating left/right bass at ~62 and ~85 Hz, still sounds like it's coming from the sides, probably due to the harmonics above the fundamental tones. JRivers analyser is a great tool to see those tones especially since they are boosted in my case.
(bass in Albatros in analyser)
Last edited:
Oh very clever. It would be clear if you showed Left only vs Right only vs combined. Why do you have to do multiple sweeps?
While I have the above screen up I might as well explain what can be seen on the left part:
Effects - Still have JRiver's surround Field on, explained it earlier, it is sort of like a Blumlein Shuffler Circuit, at least loosely based on that.
Parametric Equaliser - Also previously discussed, pre-EQ before the convolving
Convolution - DRC-fir
Parametric Equaliser 2 - Post EQ to even out the bass dip
Last edited:
I already had a lot of graphs up, else I could have shown the left and right separate ones. But what they showed was the saturation of my ground loop isolator, but it looked like 3rd order distortion in the REW graphs. A few (or many, the pages fly by lately) pages back you can see where I found out about the added distortion of that ground loop isolator. I've since moved on to optical out signal to my DAC.
Multiple sweeps in REW would be like a longer sweep, afterwards automatically averaging the sweeps and in theory getting a more realistic result by lowering the influence of the ambient noise.
Last edited:
No square waves recorded yet, the speakers are now playing as my home theatre for my kid and his friends watching Puss in Boots
I'm afraid to see what square waves would look like from my listening seat...
Since your bass boosts/cuts are not actually correcting the individual speaker response, you might want to try making them with a linear phase filter. It will add a touch of pre-echo but probably nothing to worry about. It might end up sounding slightly more transparent.
I was actually planning to record the square waves at a closer distance to one speaker.
You didn't record those impressive squares at your listening position, did you?
I assume the single speaker is going to win over line arrays for square waves. However great sound I get I do realize the time differences that will always be there from the center speaker to the outer speakers. Unless you make it a focused array.
I have a linear phase EQ plugin available to play with. I will test that in due time. It will create additional latency though. As I do the exact opposite of each other on the left/right combo I was hoping it would add up quite neutral. I'll check the phase within the DRC windowed time.
You didn't record those impressive squares at your listening position, did you?
I assume the single speaker is going to win over line arrays for square waves. However great sound I get I do realize the time differences that will always be there from the center speaker to the outer speakers. Unless you make it a focused array.
I have a linear phase EQ plugin available to play with. I will test that in due time. It will create additional latency though. As I do the exact opposite of each other on the left/right combo I was hoping it would add up quite neutral. I'll check the phase within the DRC windowed time.
Great stuff Wesayso, thanks!
I'll have to check the variable smoothing in REW. Is it in a newer version? Holm has something similar.
I'll have to check the variable smoothing in REW. Is it in a newer version? Holm has something similar.
Awaiting my damping poster cloth I have experimented with some different audio enhancers. Having a solid state setup I was wondering about the joys of tube amplification. So I played for a while with all kinds of vst plugins claiming to bring the joys of tube sound. While this was enjoyable at first, it grew old fast. One afternoon I did enjoy a plugin very much only to find it colored the sound too much the next day.
The only thing that doesn't wear out for me is the Blumlein 'like' shuffling. I decided to abandon the effect plugin from JRiver, for it's not clear to me what it does and use the Voxengo MSED plugin instead. The resultant sound is quite similar to the JRiver except now I get a seamless dial to play with.
I assume the sound/effect heard is still quite similar to the previously mentioned BSG Technologies QOL Signal Completion Stage. Only this one (Voxengo plugin) is free.
It is quite revealing to play with the plugin and mute the mid or side to see what's happening. At the right dose of "Side Gain" it makes quite a difference.
I don't think there is one setting that works for all music but so far it is quite pleasing with all the music I ran trough it.
I suspect the plugin works best if the phase behavior at the listening position is well behaved. It's nothing more than a Stereo enhancement obviously, making things like ambience, reverb and effects in the recording a little louder. But if it's pleasing, why not? It sounds more like real music to me. As it would sound like had I been there. More defined space, more depth...
And I can always turn it off at will. Actually, for Home Theatre I use the same MSED plugin to boost the "Mid Gain" a tiny bit, set "Side Gain" back to zero and pan Mid and Side to compensate for my offset TV.
I do realize I'm messing up the mix as intended by the artist. But to replicate his/her sound I would need to have been there in the recording studio to know what it should sound like.
The main thing I like about the mid/side processing is the depth I'm getting. And the "body" of instruments and singers. The background choirs are awesome. Separation of instruments is better. Well... it seems I like everything it does.
I'll revisit all settings once I have more damping in place. The damping will be added to see if I can get away with less processing. I'm quite convinced that less processing needed equals better sound. Right now I have one big damping panel, two more will be added at crucial places.
The only thing that doesn't wear out for me is the Blumlein 'like' shuffling. I decided to abandon the effect plugin from JRiver, for it's not clear to me what it does and use the Voxengo MSED plugin instead. The resultant sound is quite similar to the JRiver except now I get a seamless dial to play with.
I assume the sound/effect heard is still quite similar to the previously mentioned BSG Technologies QOL Signal Completion Stage. Only this one (Voxengo plugin) is free.
It is quite revealing to play with the plugin and mute the mid or side to see what's happening. At the right dose of "Side Gain" it makes quite a difference.
I don't think there is one setting that works for all music but so far it is quite pleasing with all the music I ran trough it.
I suspect the plugin works best if the phase behavior at the listening position is well behaved. It's nothing more than a Stereo enhancement obviously, making things like ambience, reverb and effects in the recording a little louder. But if it's pleasing, why not? It sounds more like real music to me. As it would sound like had I been there. More defined space, more depth...
And I can always turn it off at will. Actually, for Home Theatre I use the same MSED plugin to boost the "Mid Gain" a tiny bit, set "Side Gain" back to zero and pan Mid and Side to compensate for my offset TV.
I do realize I'm messing up the mix as intended by the artist. But to replicate his/her sound I would need to have been there in the recording studio to know what it should sound like.
The main thing I like about the mid/side processing is the depth I'm getting. And the "body" of instruments and singers. The background choirs are awesome. Separation of instruments is better. Well... it seems I like everything it does
.I'll revisit all settings once I have more damping in place. The damping will be added to see if I can get away with less processing. I'm quite convinced that less processing needed equals better sound. Right now I have one big damping panel, two more will be added at crucial places.
Agree it's difficult to know what the reference was.
If you can tap the unprocessed signal into headphones in real time and A/B listening taking on/off headphones this can maybe compare if plugins is coloring/distorting too much.
Think AC/DC Highway to hell album is good at this exercise because tracks are so dry (no reverb effects and nearly no ambience) that think one can hear the damped room
in the studio when listening in headphones.
I agree that headphones can supply a sensibele reference. That's why I bought a decent set when I started tuning the sound in my Car. I have the German Maestro GMP 8.35 D for that purpose. It is used as a reference set by multiple Car Audio organisations like the European Emma. I'll try AC/DC, got to have it some where.
I've asked for the secret ingredients of the Surround Field effect from Jriver but was told that they don't supply that kind of information.
I'll run some tests soon to see if I can trace what it does and compare that to what I can get out of the Vogengo MSED plugin. If the results are as similar as I think they are it would give me more control to use the Voxengo plugin.
I also played briefly with an Ambiophonic plugin I bought a long time ago. Though I'm not prepared to move the arrays closer together. So far it didn't work that well, probably due to the setup not being ideal for this kind of thing. I wish we could play with bacch-sp!
I'll run some tests soon to see if I can trace what it does and compare that to what I can get out of the Vogengo MSED plugin. If the results are as similar as I think they are it would give me more control to use the Voxengo plugin.
I also played briefly with an Ambiophonic plugin I bought a long time ago. Though I'm not prepared to move the arrays closer together. So far it didn't work that well, probably due to the setup not being ideal for this kind of thing. I wish we could play with bacch-sp!
Some thoughts about phase behaviour and speakers, just my personal view on this;
One might wonder why I'm so busy with that Jriver Surround Field effect. Especially while there is such an interesting thread going on with objective measurements of all kind of small drivers including the Vifa I used. And all the hints that we can't measure things like DDR response in speakers. Did I choose the wrong driver as a base for my array?
My motivation is that phase plays a significant role in the way we perceive sound. What gave me that opinion you may ask? Well before I was busy building these arrays I was working on getting good to great sound in my car. On the forum diymobileaudio.com there used to be a regular there with a very good handle of a lot of things important to sound reproduction by the name of Lycan.
Here's one of his threads:
midbass arrays revisited - Car Audio | DiyMobileAudio.com | Car Stereo Forum
(most of his threads/posts are of interest, also his older work under a different handle: Werewolf)
The basic principals as presented by Lycan in this thread explains a lot of the important parts of how we perceive sound.
If you think of phase as a tool to create directional queues it starts to make more sense. What I'm doing with the Blumlein Shuffling is exaggerating the phase differences already present in the recording in the left and right channel.
Without it my speakers image outside the boundaries of the speakers. But with it, in light proportions (I boost the side from the mid-side signal about ~1.6 dB), it starts to sound more realistic. Some recordings already have that boost in the mixed sound. And you have to be careful not to make it sound like a gimmick. Yet a stand alone DSP called the "BSG QOL Signal Completion Stage" gets many good reviews in all kinds of Audiophile media. It basically does the same processing, but at a fixed position.
Too much will make everything sound somewhat hollow, Exaggerating the reverb etc.
I realise that I might need less of it with free standing speakers. As the close boundaries of walls in my setup mess up the phase response of the speakers by adding different "vectors" and thus changing phase of that signal.
This is the reason I went for DRC processing. It straightens out the phase of the first wave front.
Without flat phase you can still have good imaging speakers. As long as the left and right speaker have similar enough phase response you still get the phase differences between the two channels to be able to steer the sound.
But having the phase behave more linear makes these queues more consistent to reveal what was recorded.
Actually, a bad crossover (looked at from a phase point of view) can have a rather pleasing result. You have two different sources of sound within the same speaker, overlapping in the crossover region and creating a phase difference. That in itself can sound more spacious. Except that this spacious effect isn't part of the original recording. I know I have heard that in my car, where I use active crossovers and played with numerous settings. It generally works well on recording "X" and does nothing on recording "Y".
I'm still convinced that good speakers can play everything well.
End of rant
One might wonder why I'm so busy with that Jriver Surround Field effect. Especially while there is such an interesting thread going on with objective measurements of all kind of small drivers including the Vifa I used. And all the hints that we can't measure things like DDR response in speakers. Did I choose the wrong driver as a base for my array?
My motivation is that phase plays a significant role in the way we perceive sound. What gave me that opinion you may ask? Well before I was busy building these arrays I was working on getting good to great sound in my car. On the forum diymobileaudio.com there used to be a regular there with a very good handle of a lot of things important to sound reproduction by the name of Lycan.
Here's one of his threads:
midbass arrays revisited - Car Audio | DiyMobileAudio.com | Car Stereo Forum
(most of his threads/posts are of interest, also his older work under a different handle: Werewolf)
The basic principals as presented by Lycan in this thread explains a lot of the important parts of how we perceive sound.
If you think of phase as a tool to create directional queues it starts to make more sense. What I'm doing with the Blumlein Shuffling is exaggerating the phase differences already present in the recording in the left and right channel.
Without it my speakers image outside the boundaries of the speakers. But with it, in light proportions (I boost the side from the mid-side signal about ~1.6 dB), it starts to sound more realistic. Some recordings already have that boost in the mixed sound. And you have to be careful not to make it sound like a gimmick. Yet a stand alone DSP called the "BSG QOL Signal Completion Stage" gets many good reviews in all kinds of Audiophile media. It basically does the same processing, but at a fixed position.
Too much will make everything sound somewhat hollow, Exaggerating the reverb etc.
I realise that I might need less of it with free standing speakers. As the close boundaries of walls in my setup mess up the phase response of the speakers by adding different "vectors" and thus changing phase of that signal.
This is the reason I went for DRC processing. It straightens out the phase of the first wave front.
Without flat phase you can still have good imaging speakers. As long as the left and right speaker have similar enough phase response you still get the phase differences between the two channels to be able to steer the sound.
But having the phase behave more linear makes these queues more consistent to reveal what was recorded.
Actually, a bad crossover (looked at from a phase point of view) can have a rather pleasing result. You have two different sources of sound within the same speaker, overlapping in the crossover region and creating a phase difference. That in itself can sound more spacious. Except that this spacious effect isn't part of the original recording. I know I have heard that in my car, where I use active crossovers and played with numerous settings. It generally works well on recording "X" and does nothing on recording "Y".
I'm still convinced that good speakers can play everything well.
End of rant
Last edited: