Demande d'aide pour l'interpretation des mesures acoustiques de la pièce et comment apporter les corrections

[audiolg]

Bonjour à tous,
Je suis débutant dans les mesures acoustiques et après de nombreuses recherches, tutos et en passant d’un forum à l’autre, on m’a conseillé de venir ici où les sujets concernant les analyses acoustiques et l’utilisation du logiciel REW sont nombreux. Mes questions sont d’un débutant et ce qui semble évident pour la plupart d’entre vous, ne l’est pas forcément pour moi qui suis novice dans ce domaine, c’est pourquoi, je vous remercie par avance pour votre compréhension et pour votre patiente.

Mon but est de résoudre au mieux les problèmes acoustiques de mon petit salon (4.5m x 4.5m x 2.4m) en corrigeant les fréquences qui posent problème avec un EQ. La pièce a déjà été traitée avec des panneaux et des tableaux décoratifs acoustiques, sans exagérer car esthétiquement, ça doit rester un salon.

J’utilise le matériel suivant :
PC portable (Windows 10)
REW V5.20.13
Interface audio Focusrite Solo, calibré et fichier dans REW
Micro de mesure Sonaworks avec fichier de calibrage fourni par le fabriquant.

Préampli. Rotel
Ampli. Rotel
JBL L 112
EQ numérique IMG DEQ 230

J’ai effectué mes premières mesures en un seul point d’écoute idéal en respectant la position du micro de mesure à hauteur des oreilles et des enceintes selon le triangle équilatéral.
Ci-après, les courbes que j’ai obtenues. Je suis impatient de lire vos commentaires sur l’interprétation des ces courbes et ça serait super si quelqu’un pouvait m’aider à configurer en détail toutes les options de réglage pour effectuer une analyse optimale dans la fenêtre RTA ainsi que dans la fenêtre EQ dans REW pour savoir quelles sont les fréquences à corriger.

 

[dcibel]

Using translator to read your post, I hope you can do the same for mine.

Single point measurements for room response are problematic. Average of multiple measurements across listening area is useful to capture room average response rather than single point in space. One method is "moving mic measurement" method using RTA continuous averaging and periodic noise generator to record the room response, simply move the mic in circles around the listening area, keeping constant pace of movement for averaging. Here are some settings to get you started with RTA for room EQ:

Alternative is to take many measurements (more the merrier). Apply frequency dependent windowing with default 15 cycle width. With latest 5.20.14 early access version of REW, use all SPL "cross corr align" to align the measurements, then vector average to create the average result. Personally, I think the RTA moving mic method is simpler with possibly better results.

Other item is the y axis scale in your charts provided is quite condensed. First 2 especially, 50dB/div is useless. In the preferences - view tab you will find "preset aspect ratio", I recommend 25dB/dec set here. Then you can use the buttons in the bottom right of the chart "20...20000" to set x and y axis scaling to a reasonable range.

 

[dcibel]

Other item for room EQ would be to apply phycoacoustic smoothing to avoid over-compensating every little wiggle.

 

[audiolg]

Bonjour dcibel,

Tout d’abord merci beaucoup pour ta réponse rapide.

J’ai entendu parler de la méthode du micro mobile RTA.
Personnellement j’ai l’habitude d’écouter ma musique en un seul point d’écoute, mais je vais suivre ton conseil , pour la comparaison, je vais aussi effectuer des mesures dans une zone d’écoute, je ne sais pas s’il y aura une grande différence.

Je n’ai pas trouvé le téléchargement REW V5.20.14 en accès anticipé, j’ai la version REW V5.20.13.
J’ai configuré les paramètres selon tes instructions et lorsque j’aurai effectué les mesures RTA,
je reviendrai sur ce Forum.
Salutations

 

[FargateOne]

@audiolg j'utilise DeepL https://www.deepl.com/translator (gratuit) pour être certain d'avoir des réponses...

 

[ddude003]

Welcome @audiolg, here is a link to the early access page... https://www.avnirvana.com/threads/v5-20-14-early-access-build.11154/ Post mdat files instead of screen shots as soon as you are able...

 

[audiolg]

Hello everyone,
Here are finally my first RTA measurements.
Thanks to dcbel, FargateOne and ddude003 for their advice.
I made several measurements in the listening area and here is the average curve.
1. What do you think of this curve?
2. Now, how should I proceed to configure and use the EQ window and EQ Filters.

Thanks in advance for your help

 

[dcibel]

You have many average measurements showing quite close agreement which is good to see. For a

To confirm, generator signal used was pink noise or white noise? Shape of response appears closer to pink noise at -3dB/oct. If you have used white noise, I would be worried about the lack of high frequency energy. Otherwise the curve is quite typical for in-room response.

There are differing opinions on what is the correct filter target, so recommend trying a few different settings and see what you prefer. With white periodic noise, I have found that a simple -0.7dB/oct target from 100Hz to 10kHz works well for me.

 

[ddude003]

Do you plan on using the IMG DEQ 230 for your EQ/Signal processing?

Hmm... I wonder if the "new" Generic EQ / Configurable PEQ can manage:
30 equalizers in each output
High-shelf, low-shelf and bell filters
1/24 octave steps
Gain filter: ± 15 dB each
Variable Q factor of the filters: 0.05 to 3 in 0.05 steps...???

 

[John Mulcahy]

Hmm... I wonder if the "new" Generic EQ / Configurable PEQ can manage:
30 equalizers in each output
High-shelf, low-shelf and bell filters
1/24 octave steps
Gain filter: ± 15 dB each
Variable Q factor of the filters: 0.05 to 3 in 0.05 steps...???

The configurable EQ can support the number of filters and parameter ranges, though only for the peaking filters, it doesn't offer any other types. Frequency steps are only in Hz though.

 

[audiolg]

Do you plan on using the IMG DEQ 230 for your EQ/Signal processing?

Hmm... I wonder if the "new" Generic EQ / Configurable PEQ can manage:
30 equalizers in each output
High-shelf, low-shelf and bell filters
1/24 octave steps
Gain filter: ± 15 dB each
Variable Q factor of the filters: 0.05 to 3 in 0.05 steps...???

Hello,
Yes, I have an EQ IMG DEQ 230 with which I wish to make my adjustments.
(When I know how to do it)

 

[audiolg]

You have many average measurements showing quite close agreement which is good to see. For a

To confirm, generator signal used was pink noise or white noise? Shape of response appears closer to pink noise at -3dB/oct. If you have used white noise, I would be worried about the lack of high frequency energy. Otherwise the curve is quite typical for in-room response.

There are differing opinions on what is the correct filter target, so recommend trying a few different settings and see what you prefer. With white periodic noise, I have found that a simple -0.7dB/oct target from 100Hz to 10kHz works well for me.

Hello,
Yes, I used the pink noise of REW.

 

[audiolg]

Hello everyone,
I don't know what you think but wouldn't it be interesting to make the EQ settings in two phases for comparison?
Phase 1 : First try to get a frequency response as flat as possible from 20Hz to 20kHz
Phase 2: Obtain a target curve that suits my small listening room (4.5 m x 4.5 m x 2.4 m) and especially according to my personal sound perception.

I see that you are all experts in this field.
As I have already mentioned, I am a novice in the technique of acoustic measurements, unlike most of you, I do not have enough theoretical knowledge in this field, I am simply a music lover who wants to listen to my music in the best possible conditions.
I have already been able to perform RTA measurements, this is already a good starting point for me, with your help I want to continue in my "adventure". Therefore, in relation to the result of the RTA measurement that I made, concretely, how should I proceed now with the EQ window and then with EQ Fiters, thank you for explaining me simply step by step with an example.

 

[dcibel]

Hello everyone,
I don't know what you think but wouldn't it be interesting to make the EQ settings in two phases for comparison?
Phase 1 : First try to get a frequency response as flat as possible from 20Hz to 20kHz
Phase 2: Obtain a target curve that suits my small listening room (4.5 m x 4.5 m x 2.4 m) and especially according to my personal sound perception.

First is to create average measurement, follow same process as you have previously but with periodic white noise. Phase 1 may be fairly redundant, flat in-room response is rarely preferred, perhaps only to prove that your EQ process works. I would start with target curve of -0.7dB/octave from 100Hz to 10kHz and go from there. Determining for yourself what is "better" or "best" from listening can be difficult without the ability for immediate A/B switch, so I would set up EQ and use it for some time to get a good feel for the acoustic balance. The power of manual EQ process is the ability to make any adjustment as you desire, the automation is only to improve efficiency to get 90% of the way with little effort.

I see that you are all experts in this field.

Not expert, experienced amateur

As I have already mentioned, I am a novice in the technique of acoustic measurements, unlike most of you, I do not have enough theoretical knowledge in this field, I am simply a music lover who wants to listen to my music in the best possible conditions.
I have already been able to perform RTA measurements, this is already a good starting point for me, with your help I want to continue in my "adventure". Therefore, in relation to the result of the RTA measurement that I made, concretely, how should I proceed now with the EQ window and then with EQ Fiters, thank you for explaining me simply step by step with an example.

You are on the right track. For the -0.7db/oct filter target mentioned above, here's a screenshot to get started:

Then, "calculate target level from response" will adjust the target to be centered on the measured response. Below in the "Filter Tasks", you have frequency limits to limit range for EQ. With large response adjustment sometime manual EQ band is needed, so enter manually and uncheck "auto EQ" box. Other settings here in filter task relate to your EQ hardware, gain structure, and aggressiveness of EQ automation. Play around a bit here and "match response to target" for automated process.

 

[audiolg]

First is to create average measurement, follow same process as you have previously but with periodic white noise

Hello
Why use periodic white noise instead of pink noise?

 

[dcibel]

Hello
Why use periodic white noise instead of pink noise?

For "flat" result. White noise will have the same energy across entire spectrum, while pink noise will be at -3dB/oct, so "EQ flat" with pink noise will be at -3dB/oct slope. White noise just makes it easier to view and work with.

 

[audiolg]

For "flat" result. White noise will have the same energy across entire spectrum, while pink noise will be at -3dB/oct, so "EQ flat" with pink noise will be at -3dB/oct slope. White noise just makes it easier to view and work with.

ok I understand.
So I'm going to do some more measurements and the average RTA with white noise.
I will follow your instructions for configurations and when I am done I will post the results.

Thanks and see you soon

 

[John Mulcahy]

For "flat" result. White noise will have the same energy across entire spectrum, while pink noise will be at -3dB/oct, so "EQ flat" with pink noise will be at -3dB/oct slope. White noise just makes it easier to view and work with.

Only if you are using the RTA in spectrum mode rather than fractional octave mode. White noise means disproportionate energy at HF, opposite to ambient noise characteristics and the power handling capabilities of drivers. White noise has its uses for electronic measurements but I wouldn't use it for acoustic measurements, better to use pink noise and use the RTA in fractional octave mode.

 

[dcibel]

For most room EQ scenario, power level will be maybe 1W output, so power handling is irrelevant. Sine sweep measurement has similar relation of power vs freqency as white noise. One other option I could think for pink noise would be for +3dB/oct correction to flatten the curve for visual appearance of EQ.

 

[dcibel]

With some trace arithmetic, and using EQ target to generate a 3dB/oct upward slope, Pink noise measurement can be corrected to match the white noise results, at least above 40Hz in my quick comparison.

Other option is RTA instead of spectrum mode, it can look very similar, here I did pink noise RTA at 1/48oct followed by psycho-acoustic smoothing. In screenshot below you can see 3 methods of generating very similar results in green, yellow, and red.

 

[John Mulcahy]

Sine sweep measurement has similar relation of power vs freqency as white noise.

Log sine sweeps have a pink profile.

 

[dcibel]

Log sine sweeps have a pink profile.

Interesting, for some reason I was under the impression that sine sweep would be equal amplitude over the entire spectrum, not decreasing in amplitude at 3dB/oct slope.

 

[John Mulcahy]

It is equal amplitude, but it spends equal time in each octave so its spectrum drops at 3 dB per octave.

 

[dcibel]

I guess "spectrum" mode in REW just does not operate as I would expect. Spectrum analyzer in ARTA operates same as RTA mode in REW, pink noise loopback provides flat response. To be sure I was comparing the same signal, I used ARTA spectrum analyzer with external signal, and file saved from REW generator, same result.

Comparison below, same signal, same y axis units, same FFT length, window type, and exponential averaging used.

In any case, I had used white noise with REW in the past successfully, I guess just don't blast that noise too loud to not stress your tweeters.

 

[audiolg]

Hello
I am very impressed by your knowledge but
I see that there are different opinions on this issue.
I prefer not to intervene in your discussion because I don't have the knowledge.
Without offending any of you but to finish, what do I do ???
Pink noise or white noise ?