Dirac Live newbie... anyone else's room measure so badly...? Is bass "boost" so wrong...?

[eternaloptimist]

Hi Everyone,

Have been starting with Dirac Live, stereo....

Room is large, rectangular, with speakers pointing down the long axis, with one speaker in a corner too near the side wall (mostly glass...) and the other 2.5m apart and far away from a side wall (cannot be changed, unfortunately...). High 4m, uneven ceiling.

Listening position is 3.5m from the centrepoint of the speakers.

Speakers have AMT tweeters.

Used "couch" setting for measurement and UMIK-1 microphone.

 

So.... this is the waveform, both left and right. There is a large spike at 70Hz on the left (?!)

 

The suggested target curve.... I found not very dynamic and "flattened"....

 

To get some of the bass back...! I have found this much better.... Large sweet spot, better imaging, lovely soundstage!

 

I am not sure if I have committed audiophile blasphemy with the target curve....

I would welcome any comments or suggestions!

[wgscott]

Everything you have looks completely normal. I tried a bunch of target curves, and stayed with their House curve and one with a bass hump like you have.

[LBob]

I don't think your room measures that badly. Mine was way worse. I'm using pretty much the same curve that you are: slight elevation in the bass and a roll off for the high end. Enjoy!

[tne]

Pretty much any room that is not a dedicated music room designed as such, is pretty bad. It is just that now you have made measurements and have a graphical representation of the response function. As Bill and LBob stated, this all seems normal.

 

Plus, if it sounds good, it is good.

[eternaloptimist]

Thanks for the comments, guys!

I'm also experimenting with just correcting below 300Hz; leaves the 2-4 kHz dip which I quite like (or am just used to!).

I have found I have had to adjust the gain slightly to "move" the balance to the right speaker (-1.6 db off the left). I wonder if I need to remeasure and ensure the sweet spot mic position is accurate....

All in all, this is quite fun!

The imaging is certainly sharper.

[flak]

Thanks for the comments, guys!

I'm also experimenting with just correcting below 300Hz; leaves the 2-4 kHz dip which I quite like (or am just used to!).

I have found I have had to adjust the gain slightly to "move" the balance to the right speaker (-1.6 db off the left). I wonder if I need to remeasure and ensure the sweet spot mic position is accurate....

All in all, this is quite fun!

The imaging is certainly sharper.

 

Yes, the dip around 3 KHz is also called "BBC DIP" and is discussed for example here:

Electro-acoustic models

there is no reason why you cannot incorporate it in your target

 

I have some reservations instead about modifying the gain of the right speaker as computed by Dirac Live because it normally is the right one...

you can double check that by either using your system in mono (if you have that possibility) or by playing a monophonic test signal that you can find online, you should find out that the source is perfectly centered

(this is the first site that I've found: Download Audio Tone Files)

 

Flavio

[eternaloptimist]

Thanks Flavio, very interesting.

I suspect I may have not measured the sweet spot in the correct position....i.e. off to the side....

I will try and remeasure in the next couple of weeks.

I will also try and set a target curve with this 3 kHz dip.

Dirac Live has produced great results!

 

I was wondering....

I use a Macmini bootcamped with Windows and generally use JRiver on Windows.

With the Dirac live stereo licence, is it permissible to install also on the OS X partition, on the same computer? I ask, as I also use Audirvana. If so.... Would I be able to request access to the beta Plugin creation tool so the filters will work with Audirvana?

[flak]

Thanks Flavio, very interesting.

I suspect I may have not measured the sweet spot in the correct position....i.e. off to the side....

I will try and remeasure in the next couple of weeks.

I will also try and set a target curve with this 3 kHz dip.

Dirac Live has produced great results!

 

I'm pleased to read about your great results and I agree that it is possible that the measurement may have been taken at the center of the speakers instead of the main listening point even if asymmetrical

 

I was wondering....

I use a Macmini bootcamped with Windows and generally use JRiver on Windows.

With the Dirac live stereo licence, is it permissible to install also on the OS X partition, on the same computer? I ask, as I also use Audirvana. If so.... Would I be able to request access to the beta Plugin creation tool so the filters will work with Audirvana?

 

I've not tried using a Mac bootcamped with Windows on the same machine as I see no reason to do it... the license is now valid for both Mac and Windows so you can use both Audirvana and JRiver on your Mac with Dirac Live without Bootcamp...

and yes, you can have the beta plugins, please send me a PM.

 

Flavio

[Kafy]

Hello,

 

I am a newbie in the room correction too.

From your pictures I would not boost the low frequencies as you try (from 15hz to 30Hz).

I think this could create some distortions, because you force against the normal speakers response.

But I am maybe wrong.

Does any one can confirm it ?

 

From my experience, flat response is not the correct one to go because in fact you don't know if the record was made with a flat response. I think it is better to make the ones you like for differents kind of music.

 

This could be done with a standard Equalizer but what Dirac Live do well is a good channel equalization and temporal alignement and it improve the impulse response.

[KnockKnock]

Dear eternaloptimist

 

Don´t worry, your measurements look quite good, above average I´d say.

 

Try to follow the measurement with your Target curve, then you won´t loose the characteristics of your speakers that you love ;=)

 

You can see my example here below , I just follow the measurement curve with the target curve. And don´t make a new target curve for every song you have, make the same target curve with 3-4 different more/less bass, so you can easily switch between them ;=)

 

Enjoy and have fun

 

 

KnockKnock

Dyn + Bass.pdf

[eternaloptimist]

Dear eternaloptimist

 

Don´t worry, your measurements look quite good, above average I´d say.

 

Try to follow the measurement with your Target curve, then you won´t loose the characteristics of your speakers that you love ;=)

 

You can see my example here below , I just follow the measurement curve with the target curve. And don´t make a new target curve for every song you have, make the same target curve with 3-4 different more/less bass, so you can easily switch between them ;=)

 

Enjoy and have fun

 

[ATTACH=CONFIG]22751[/ATTACH]

 

KnockKnock

 

Thanks KnockKnock,

I will have try doing as you suggest and report back!

You are right... the Audiovector speakers I have sound great and I don't want to lose that.

[eternaloptimist]

I'm pleased to read about your great results and I agree that it is possible that the measurement may have been taken at the center of the speakers instead of the main listening point even if asymmetrical

 

 

 

I've not tried using a Mac bootcamped with Windows on the same machine as I see no reason to do it... the license is now valid for both Mac and Windows so you can use both Audirvana and JRiver on your Mac with Dirac Live without Bootcamp...

and yes, you can have the beta plugins, please send me a PM.

 

Flavio

 

Many thanks for the info and the beta plugins!

I will send a PM.

 

Dave

[jjkale]

Hey KnockKnock,

Happy Holidays. jjk43 here.

I like the curve you designed. I am about to try my first curve this weekend.

Your screenshot didn't include the db levels. Did you lower the bass shelf to 0db and the remainder to - levels? If not , just wondering about any clipping. If so, how do you lower an entire curve by say 8db?

Also did you set distance levels and trims in Jriver (or other) prior to running Dirac? The DL manual only mentions setting the number of speakers correctly.

Thanks very much, good weekend

jjk43

[KnockKnock]

Hi jjk43

 

Merry Xmas to you too

 

My advise is just to follow the measurement curve, and rise the lows maybe 6-8 dB. If there is a dip about eg 1 kHz, follow it with your target curve.

I'll slightly roll of the top end from say 1-2 kHz with a slow (not parametric) curve.

When all this is done save your target curve (or more with eg more or less bass), close the App, and open DiracLive. There you can adjust the final level (click on the very small arrow), I've set my level to minus 4 dB so no overloading is done

 

best KnockKnock

 

Hey KnockKnock,

Happy Holidays. jjk43 here.

I like the curve you designed. I am about to try my first curve this weekend.

Your screenshot didn't include the db levels. Did you lower the bass shelf to 0db and the remainder to - levels? If not , just wondering about any clipping. If so, how do you lower an entire curve by say 8db?

Also did you set distance levels and trims in Jriver (or other) prior to running Dirac? The DL manual only mentions setting the number of speakers correctly.

Thanks very much, good weekend

jjk43

[wgscott]

My advise is just to follow the measurement curve, and rise the lows maybe 6-8 dB. If there is a dip about eg 1 kHz, follow it with your target curve.

 

Why do that? Doesn't that just defeat the purpose of room correction?

[Kafy]

In my opinion, the flat target is not the correct one to go :

 

 

Because :

1 - The flat target is the target at the output of the speakers. The flat target just means that the sound at the sweet spot is the same as the sound from output of the speaker.

Try to listen your speakers each at 1m of your ears and you will see it sound awfull.

This could be good for monitor speakers but our speakers have been made to be use at 2 to 4 meters.

 

 

2- The flat target means the same air pressure at any frequencies. Does this spot exist during the recording ? where it is ? for me there is no spot with a flat response in a concert hall or in a recording studio.

 

 

3- What could have a flat response in the concert hall or recording studio ? only the microphones. To be sure there is no sound modification done by the microphone. So, what we expect is that the speakers can reproduce what the microphone has recorded and this is why we expect the speakers have a flat response at the output, but not at the sweet spot ?

 

 

What should be the target to the sweet spot ? For me it is very diffucult to identify it because it depends of :

 

 

1- What the speakers manufacturer have done, and this is the sound you like and the reason you have bought them.

 

 

2- What your room sound ?

 

 

3- what the engineer have made during the recording; all records are differents.

 

 

 

For me the correct target is the one you have measured and you just have to modify some parts to remove some reseanonces. This can only be done with your ears or may be a software can do it but I don't know it.

If this kind of software exists, it must be able to get all specifications of recording room (size, materials...) and your own room to be able to create a specific correction to re-create the original sound (from the concert hall) into your room ...... this is not possible for the moment because the informations of the recording room doesn't exist in the recording.

 

 

 

For the moment the room correction is only the measurement and the correction but not the creation of the target which is to complicated to do because of the missing informations about the original recording room.

 

 

 

 

This is only my opinion :-)

[Kafy]

And I've forgot to say that does our listening has a flat curve ? I mean does a flat curve at the listening position will be heard as flat ?

I don't think so. All I have read about the psycho-acoustic show the human hearing is not flat. So until we have microphone with same response curve than the human hearing, there is no need to try to get a flat curve at the listening position.

[wgscott]

So what makes you think the irregularities are due to the speaker rather than the room? Why would you even bother with room correction software?

[Fitzcaraldo215]

And I've forgot to say that does our listening has a flat curve ? I mean does a flat curve at the listening position will be heard as flat ?

I don't think so. All I have read about the psycho-acoustic show the human hearing is not flat. So until we have microphone with same response curve than the human hearing, there is no need to try to get a flat curve at the listening position.

 

 

Who said anything about a flat curve? Dirac's stock curve is not flat, rather it is downward sloping from low to high frequencies. That is based on extensive research and testing from a number of sources to give sound that is perceptually flat to the ear. A truly flat curve where the level of the highs is equal to the mids and lows sounds too bright to most people.

 

Different researchers have come up with very slightly different recommendations, but they all agree on the smooth downward slope. The Dirac target curve is almost identical to the one developed by instrumentation microphone maker B&K. So, Dirac is not advocating flat.

 

You are, of course, free to fiddle with the curve however you want. After hearing many such fiddlings and tweakings, I am quite happy with the stock Dirac target curve.

[wgscott]

I was also unable to improve on the Dirac in-house stock curve. I did make a few others for experimentation.

 

I am very impressed by the competence of the Dirac folks and their software, and think their recommendations are well thought out and worthy of following.

[deeejeee]

I think te flat curve is just right at the volume you have made measure.

If you listning at a louder volume it seems only the mid and high change.

[Kafy]

First of all, english is not my natural language, so I maybe not understand and explain right. Sorry about that.

 

 

 

 

To wgscott :

I have nothing against the room correction software, I am an happy user of Dirac Live.

I am aware that many of the issue (like bases resonances) are due to the room.

I don't say the irregularities are due to the speaker, but each speakers have it's own properties. Even if all speakers have now all a very good flat response at 1m, each speakers have their own properties that make them sounding dirrerents.

I am sur that in the very near future this will not still exist because of the abilities to modify and adapt the entry source to the room and the speakers.

See the Devialet SAM that modify the entry signal to match the speakers ability.

 

 

 

 

 

 

To Fitzcaraldo215 :

 

 

I didn't explain well.

When I talk about flat response, I mean it is straight from low to hight frequencies.

I know there is a slop but it is still "flat", there is no "mountain".

 

 

I am not a specialist and I am still learning and for the moment during ma search I have never found a definitive answer about the target curve.

 

 

There is a lot of curve, all made after many research.

 

 

Some people prefer to use the ISO 2969 target curve, some a flat and slop curve.

Some want to use isophonics curve which have been modified the last years.

 

 

 

 

That means, this question of target curve is not still completed.

 

 

 

 

Here is a very interesting test to compare the subjective and objective preferences :

https://docs.google.com/file/d/0B97zTRsdcJTfY2U4ODhiZmUtNDEyNC00ZDcyLWEzZTAtMGJiODQ1ZTUxMGQ4/edit?hl=en_US&pli=1

 

 

For me, that let me thinking that if the objective curve (made with measurements) is not the prefered, that means this curve is not the right one to go and there is something this curve has forget.

 

 

The isophonic curve seems to show that the human hearing has not a flat sensibility.

https://en.wikipedia.org/wiki/Fletcher–Munson_curves

Our hearing is not flat as a microphone so why would we create a flat curve a the listening position ?

 

 

The most important is to not forget that our hearing know the answer. Our hearing has learn during the human history to identify what is the sound, where is it located and is it a normal sound.

 

 

I often compare this with the a 3D digital face. Even if it is very well done, you will easy identify it is not a real face. Because your eyes have learned to identify the emotions in face during human history.

 

 

 

 

Again, I am not a specialist, so if any one is able to explain the correct target to go I will be very happy.

For the moment, the default Dirac target is not sounding correct to me and I use a target that follow the curve of my speakers original with some corrections in the low frequencies.

[Fitzcaraldo215]

I wish you luck in your search for the perfect target curve. You will need it.

 

First, let's look at the Sean Olive study of listener preferences on page 24:

 

https://docs.google.com/file/d/0B97zTRsdcJTfY2U4ODhiZmUtNDEyNC00ZDcyLWEzZTAtMGJiODQ1ZTUxMGQ4/edit?hl=en_US&pli=1

 

That shows frequency response at the primary listening position, and the top curve was preferred by the listeners.

 

That curve is very close to the standard Dirac target curve and to one developed by B&K. The roll off slopes might differ very slightly, but they all are essentially the same idea. So, Olive's research has independently confirmed a listener preference for a smooth, downward sloping curve very similar to the one used by Dirac.

 

Second, you referred to ISO 2969, which is the famous X-Curve used in movies. I do not think you will get good results from that because it was derived for large spaces, like movie theaters. That curve goes back to speakers used in theaters in the 1930's and it has many critics today among engineers and acousticians.

 

Fletcher-Munson frequency sensitivity adds complications, and most of this research target curve assumes a "reference level" that is fairly high. You could create additional Dirac targets with varying degrees of deep bass and treble boost to compensate for this, selecting them when you listen more softly. But, I find we tend to mentally compensate when listening to soft music for the decreases in our hearing sensitivity at the frequency extremes. So, I have not found that worth the time. I am fortunate, perhaps, in having a listening room where I do not need to turn the volume down often.

 

My own sense from listening to the stock Dirac curve, and to Audyssey MultEQ XT/32 with an Audyssey Pro calibration for many years before that, is that Dirac does a great job. In my opinion, with the stock target, it delivers sound that is satisfyingly close to the live classical concerts I attend frequently.

[Kafy]

Thank you for your explainations, it helps me to understand better.

 

 

Since the last time I have tried again the default Dirac curve, and keep it for two days.

 

 

I use Dirac for less than a month and I am maybe too accustumate to how my system and room sound together.

It is like eating with too much salt for years and when you eat a normal meat, you miss the salt :-)

 

 

With the Dirac target all is well balanced, softer but it miss some air (salt). I think I have to modify it a little bit.

 

 

Thank's again for your help.

[flak]

The isophonic curve seems to show that the human hearing has not a flat sensibility.

https://en.wikipedia.org/wiki/Fletcher–Munson_curves

 

That's correct but we should take into account that that curve is already "built in", we don't need to emulate it in a Dirac target curve... what we should eventually consider is the difference in sensibility among those curves at a different listening volume.

 

Also a lack of "punch" or "impact" of the bass is a common complaint... even if it is caused by the removal of the bass boost at some frequencies caused by the room, modifying the target with one with a steeper slope or a bass boost is a good idea.

It is anyhow possible to easily modify it later on if, in time, someone finds that the bass boost should be reduced.

 

Happy New Year

Flavio