Room correction

Hi, I use Jriver but I would like to change to Audirvana.
I use online filter (room correction) in Jriver (Convolution menu).
Can I use an online filter in Audirvana also?

1 Like

I’m using the AU N-Band EQ plugin for room correction. It supports up to 16 parametric EQs.

I calculate the PEQ parameters using Room EQ Wizard, and enter them manually in the AU plugin settings popup.

1 Like

I got it. My problem that I use waw file for online room correction in Jriver.
I measure with REW the SPL in the room and REW calculate automatically a filter (I set my target curve of course also).
This filter will be exported from REW in special format (xml) to REPHASE software. I correct manually the phase of filter and the output of REPHASE will be a waw file.
This waw file will be set in the Jriver Convolution menu as an online filter (I use this type of correction only under 150Hz.)

Question to developers:
Do you plan to improve the features of Audirvana with the on-line room correction?


If you have a mac you can use LAConvolver

It works with Audirvana 3.5.33 in my 5.1 configuration

I have to select “Configuration manuel pendant la lecture (inactif pendant la lecture DSD)”
“Manual configuration during playback (inactive during DSD playback)”
And don’t use the dark mode because you can’t see well in the LAConvoler window in the dark mode.

One mono IR file per track is required.
In multichannel, the configuration is only kept for channels 1 and 2.
I must recall a preset at each stop of playing.
For DSD files I let audirvana convert to PCM of course.

On the other hand convolution restart at each tracks unlike the previous version of Audirvana Plus which keeps the convolution works between tracks.

The gain of the LAConvolver procees could clip the audio signal,
In 32 bit float (AU stream) this is not a problem except for the DAC.
I use dpMeter XT (wich shortly mute the sound every 90 seconde without licence) to controle the crest of each channel.

For Windows I don’t know an equivalent plug in but there may be in VST.
dpMeter XT is in AU and VST.

1 Like

Thank you for your detail explanation. I use Audio PC, so I need the an other solution. :slightly_smiling_face:

Perhaps Room Shaper ?

MConvolutionEZ works in VST3.
It is made for reverberation FX so there is some functions we don’t needs. It also copy a lot of IR impulse files but it is free.

I have tested with a 24 bits 48kHz neutral IR stéréo file, for a 24 bits 48 kHz stéréo source file and neutral setting (Dry/wet 100%, Widenning 0%, Higpass Off and LowPass Off) in a professional software and it seems to be working.
The only differences (around -130 dBFs peak level) are probably due to rounding errors from floating point calculations as for any processing.
Not tested in Audirvana, not tested with differents sample rate, not tested with multichannels.

I’m glad to hear it’s working for you - I’d like to do the same. How do you translate the info from REW to be manually inputted, for example, is Q the same as Width or is there a mathematical way to translate it? Ideally I could just upload a filter file but I can’t figure out how to do it.

This is how I calculate the bandwidth (in octaves) for the AU N-Band EQ, from the Q values REW provides:

BW = asinh(1/(2Q)) * 2/ln(2)

( “Relationship between Q and bandwidth”)

This is the spreadsheet I use for that:

Thanks. A friend on ASR provided this link. There are a lot of other useful tools here as well but this might save some time!

hi Gat, are you saying we should not export the REW corrections as stereo and then apply that to both/each channel in the LA Convolver? This is what I have been doing, but seem to be wrong.
Also, I struggle to manage Gain and make sure nothing clips, neither in the DAC not in the amp. I believe I have opted for the Conservative gain, to prevent any issue, but I am unsure how / if this works 
 All tips are welcome :slight_smile:

Bonjour danberilloux,
LAConvolver utilise uniquement des fichiers mono, un fichier par canal et qui ne sont pas identiques si les corrections à appliquer sont différentes.
Je ne connais pas REW mais si REW produit un fichier stéréo il faut séparer les canaux en deux fichiers mono distincts.

Vous avez tout a fait raison de vous soucier du gain et des niveaux crĂȘtes.
En fonction de la musique et des traitements appliquĂ©s il peut y avoir beaucoup de clips. Voici un post qui Ă©voque le problĂšme Control the level and the peak (crest) / ContrĂŽler le niveau et la crĂȘte max

Les signaux dont ont rĂ©alise la convolution doivent ĂȘtre de frĂ©quence identique. Aussi j’utilise systĂ©matiquement la conversion en 96kHz d’audirvana car mes fichiers de correction sont dans cette frĂ©quence. LAConvolver gĂšre normalement la conversion de frĂ©quence mais je n’ai pas testĂ© et je prĂ©fĂšre le faire faire par Audirvana.
Je place dans l’ordre dpMeterXT / LAConvolver / dpMeterXT ce qui me permet de contrĂŽler les niveau et les crĂȘtes de chaque traitement.
J’en suis arriver Ă  la conclusion que pour moi LAConvolver doit ĂȘtre rĂ©glĂ© de la maniĂšre suivante. Wet Gain -10 dB, Gain compensation Normal, Pas de Dry Gain. On perd beaucoup en niveau sonore mais c’est le prix Ă  payer pour ne pas Ă©crĂȘter.
Et pour les fichiers en DSD aucun gain supplémentaire lors de la conversion en PCM.
En fonction de la dynamique des album que j’écoute la crĂȘte max envoyĂ©e au convertisseur se trouve dans une plage allant de -10 dBFS Ă  -0,1 dBFs.
Une crĂȘte > 0 dBFs dans la chaĂźne de traitement n’entraĂźne pas de clip si tous les traitement travaillent en 32 bits flottant, mais il ne faut pas envoyer de signaux dont les crĂȘtes sont > 0 dBFs au convertisseur.
Si vous utilisez le contrĂŽle de niveau logiciel et non par le DAC cela peut Ă©galement influencer la crĂȘte max envoyĂ©e au DAC de la valeur du niveau d’Audirvana.

Merci! Is there any objective way to know when the level is too high and we have digital clipping? I clearly heard some when I ran a sweep 10 hz to 200 hz and pulled the level down, but I’d like to be more precise and also cover the full 20 to 20k range, which I can’t do by ear.

I use Sonarworks Reference for room correction with Audirvana (and Foobar2000). The VST Plugin works great. With the VST plugin you can have the room correction before WASAPI or DLNA output and therefore bypass windows mixer.

I formerly used Dirac but switched to Sonarworks.

Out of interest, why don’t you use Dirac anymore? Or what is the advantage of Sonarworks?

I purchased Dirac in 2012 und used it for many years. Then I stumbled upon Sonarworks and in comparison to the old(!) Dirac version there were several advantages, especially in the measurement process, for example more points of measurement (9 vs. 37), an individually calibrated microphone (in comparison to an average calibration file) and better guidance through the process (with click sounds which guide to to the exact spot, you can find a guy on youtube showing that) all of which results in a better measurement.

Sonarworks also includes headphone calibration in the same software, so you have both two channel and headphone calibration.

Last but not least it’s a software for professional use which automatically earns more of my trust than any audiophile gear. Their support is good as well.

Concerning sound quality I can’t compare because I moved and the rooms and gear is completely different. I’d say both do their jobs well. With Dirac you can do manual tweaking of individal frequencies, but I don’t need that. If I needed that I’d add EQuilibrium also as VST, which is a transparent EQ.

I got the support to upgrade my old Dirac license to the new version but it is so completely different that I found it too confusing and sold it instead of comparing it directly to Sonarworks Reference.

Actually Sonarworks VST might be the reason why I might never check out Roon, because Roon will never support VST plugins and I don’t know how to do room correction there then.

EDIT: You can also choose the amount of correction which is applied by Sonarworks. The old Dirac was always 100%, if I remember correctly. But 100% is most times too much. I and others have found that 70 - 80% sounds best. Again, I can’t say anything about the new Dirac version 3.x.

Hello MeduimRare,

You could use DPMeterXT for multichannel configuration

Perhaps mvMeter2 or dpMeter 5, not tested yet.

I think I will run REW at different output levels and see what distortion I can find changing.

REW v5.20 RC6 now supports the AU N-Band EQ directly, there is no longer a need to convert Q values to BW.

1 Like