New loudspeakers & room digital correction service

[Alec_eiffel]

By the way, I noticed yesterday that the most recent Roon version is processing the convolution much more efficiently than before. Earlier the processing speed indicator was always showing something between 20x-50x. Now the processing speed indicator has disappeared, meaning that it's > 100x.

Indeed, this is good news. Before the Roon update I was at 30x on a regular 44.1K/16bit track, but down to 2x on DSD64 with all the conversions and nearly a million taps of processing. Now I am at 50x on 44.1K/16bit and 5x on DSD64.
This means there is little need to design "lighter" filters for high frequencies and one can easily live with a single 44.1K set of filters, which will be automatically converted by Roon for higher frequencies (number of taps increases with frequency) at the expense of processing power but who cares.

[Hifi_swlon]

I've added a topic at Roons forums too.
https://community.roonlabs.com/t/room-co...udio/30143

Lots of Room Correction and convolution boys and girls (actually, thinking about it, I can't recall a female here or on Roons forum - bit sad really) there that may find it of interest.

[Celts88]

Hi Guys (and/or gals)

I got my HAF filters from Thierry last night, and below is the before and after graphs he sent me for my set-up.

Wouldn't have a clue what's good or bad within the graphs as I'm a complete novice at all this

Levels before correction:

   

Levels after correction:

   

Spectrogram comparison (top = w/o correction, below = with correction) for the sweet spot.

   

[sam1000]

what happens when I move my speakers a few inches or I re-arrange the furniture in the room? DO I have to take the measurement again and get new filters?

PS - I do that for my HT setup with Marantz processor which has Audyssey Room correction, but it shrinks the sweet spot quite a bit.

[homeaudiofidelity]

Regarding speakers movement up to 3 inches would be OK, for the furniture I have no generic answer as each case can be very different.
My recommendation is that your ears should be the first judge (do you experience a downgraded performance after the changes ?)
If the answer is yes you can make a measurements iteration & ask for my opinion

[sam1000]

Thanks Thierry. I'll take some measurements in next couple of days and we can take it from there. Thanks for joining the forum and answering questions.

[Vivialet]

I gave up on using convolution long time ago. From what I can see from the graphs is that you have a much better frequency response in your room, but also that some deep frequency holes are stuffed now by increasing the level. This is killing the dynamic in my optinion, you have to raise the volume by 6 or 8 db to get the same volume.
Also filters created with REW are out of phase. For the advantage of a better frequency graph you get some (for me unacceptable) downsides...
I prefer using the free mathaudio plugin in foobar to do an offline folding of my audio files, then I don't need any online processing which is also good for sq.
Mathaudio is only removing peaks and also corrects the phase. For me the best sounding correction, and I tried a lot, e.g. Dirac, REW convolution, Accourate, ...

[petrik]

Dirac, REW convolution, Accourate, ...

I have tried the same and I was never completely happy with those. The HAF is much better in my opinion. I have never tried the Mathaudio but I suppose that's not for me since I want to use Roon. Having to use Foobar or JRiver would be a showstopper for me at least

Perhaps you could try HAF and compare it with Mathaudio? I would be interested in knowing which one you prefer.

[Hifi_swlon]

I've not tried mathaudio either, but as petrik it's a non-starter as I want to use Roon and don't like the idea of pre-converting my library.

I've personally never perceived any sound quality degradation with any convolution I've tried - other than during experimentation. I'm also more than happy to attenuate the whole signal to bring everything more in line with an ideal house curve. I don't notice any lack of dynamics personally - far from it the sound is far better. I guess in my setup the boom and L-R mismatch are worse than any negative I'll get from EQ. I've got plenty of room on the amp too so more than happy to run it louder.

So far this is just talking about Dirac and REW. I've not tried Acourate, and imagine I won't now - unless I don't like Thierrys results (when I finally get round to measuring and testing his service).

[Alec_eiffel]

I gave up on using convolution long time ago. From what I can see from the graphs is that you have a much better frequency response in your room, but also that some deep frequency holes are stuffed now by increasing the level. This is killing the dynamic in my optinion, you have to raise the volume by 6 or 8 db to get the same volume.
Also filters created with REW are out of phase. For the advantage of a better frequency graph you get some (for me unacceptable) downsides...
I prefer using the free mathaudio plugin in foobar to do an offline folding of my audio files, then I don't need any online processing which is also good for sq.
Mathaudio is only removing peaks and also corrects the phase. For me the best sounding correction, and I tried a lot, e.g. Dirac, REW convolution, Accourate, ...

Filters created with REW do not need to "stuff" deep frequency holes (which is a bad thing to do fully agreed) and are not necessarily out of phase. It depends on what you ask. With the help of Rephase you can design "minimum phase" filters that prevent pre-ringing (Mathaudio probably does this automatically).

Why would an online convolution cause SQ issues versus an offline conversion ? Anyway this is a no go for me to convert files online because I stream a lot of music from the web (Tidal, radios).

Indeed it will be interesting to compare HFA with Mathaudio processing.