On-line room correction services

[AudioGeek]

I have played with lots of different DSP/DRC methods. REW, dirac, Acourate, minidsp, deqx etc.

 

For me the easiest and best results have been with using Acourate to generate convolution filters in Roon and HQP.

 

I tried HAF filters about 2 yrs ago with crosstalk reduction. I think it was cheaper back then - like < $200. Definitely the best filters I have ever had!

 

The problem with using HAF is I change components, positions, rooms, speakers quite often. Hence much more practical, cheaper to do myself.

 

If you use dirac, its really straightforward. If you want to generate convolution filters there are plenty of online guides / books for REW and Acourate.

 

No experience with DRC designer. It looks like it will do the job.

 

I completely understand people wanting to outsource dsp programming. Unless you really enjoy it, there are better ways to spend your time!

[davewantsmoore]

4 hours ago, dan_b said:

The filters Mitch developed (which run in Roon) are much better than I could manage with Dirac. I feel like I’m getting the benefits (even bass response and better soundstage) without the disadvantages I had with Dirac (a general loss of dynamics). 

You're likely already aware but it's extremely important to ensure the target curves are identical (and also confirming the overall amplitude response results are the same).     Slight changes, can result in all sorts of subjedctive differences.

 

That being said, I don't doubt your result.

[davewantsmoore]

4 hours ago, AudioGeek said:

No experience with DRC designer. It looks like it will do the job.

On it's own it doesn't come with any tools to display the results and optimise the "settings".

 

As Ric was getting at, all the tools can DO what is needed, and they're all of equivalent quality in that regard (the dsp/math) .....  the main thing which is going to affect the subjective quality is what choices are made.

 

Example:  As you're stepping through the Acourate "step by step guide" ..... the choices that you make for the windowing, and whether they're the best choice for the room/speaker.    The out of the box settings aren't necessarily at all optimal.

[Guest rmpfyf]

12 hours ago, davewantsmoore said:

You pivoted to the underlying DRC.   Yes, it can do all the things.....  but so what?

No pivot Dave - last post on this as we're derailing it - the core functionality under each of these solutions really doesn't differ significantly. My exact words were 'no less good than anything costing thousands', not 'just as good result as paying thousands'. Invest well and results should improve, that's logical, insofar as core capability with respect to DSP - no less good. I appreciate you're suggesting limited capability in the UX gives rise to limited results. I'm suggesting an online approach for many has other limits. 

 

12 hours ago, davewantsmoore said:

You're suggesting an alternative to the "professional online service" is DRC designer.

 

My comment is that it WON'T give as result anywhere near as good as the pro-service (and quite a bit of tinkering for the average-man). 

Yes, it's an alternative for said 'average-man'. Could you get a better result than with DRC Designer? Duh, yes. Could you get the same result? Worse? Also possibilities. 

 

There's a good deal of knowledge (understatement) involved in just taking accurate, representative measurements (as has been stated by others on this thread). 

 

I would wager the market for people who want a filter generated, do not want to tinker with it at all, know how to take measurements of necessary relevance/accuracy/resolution, do not want to invest in software to generate said filters and do not want a personal relationship with a person generating said filters... is small.

 

If considering as much to start with, start with the easy stuff. I may have erred in assuming the OP was looking for a place to start. 

 

Going SaaS with poor inputs can p*** away $ quickly. Burn the learning elsewhere. @dan_b has a great story - his direct cash outlay is reasonable however the time involved in taking measurements, accruing experience etc is not inconsiderable either.

 

I wouldn't suggest DRC to be a lifetime of learning either - it's specialist, but hardly a person's complete meditation on life - if anything the Denis Sbragion's writings on the topic make good reading.

 

Happy tweaking, all.

[davewantsmoore]

40 minutes ago, rmpfyf said:

There's a good deal of knowledge (understatement) involved in just taking accurate, representative measurements (as has been stated by others on this thread).

In this specific situation.... I don't agree.... at all.

 

All you need to do is put the microhone in the listening position .... observe a few simple caveats  .... and capture the output.

 

Aside from technical issues such as broken computer setups etc. (which are obvious in the results, and the service provider assist you to resolve) .... then there is zero other knowledge, decision, or consideration required from the user.

 

There is nothing to optimise in the "taking of the measurement" part.   It is simply capture from the listening position.

40 minutes ago, rmpfyf said:

know how to take measurements of necessary relevance/accuracy/resolution

See above.   What do they need to know?    Aside from following some simple instructions.  Nothing.

 

Nothing the user will be asked to do, (or decide - as they are not asked to decide anything), is going to affect the relevance, or accuracy, or resolution, of the data.

[davewantsmoore]

23 hours ago, Steffen said:

One thing I learned along the way is to keep EQ out of the higher frequencies, and to limit it to the lower few hundred Hz.

Yes, this is the key problem.

 

When you measure the response from the listening position.... the issues becomes understanding what (time) part of the response to correct and by how much.    It is safer to correct the LF (as you noticed) ..... but the majority of the audio (eg. 100 to 1000 Hz) is in a big band of "it depends".... and this is where much of the analysis/judgement comes in.

[AudioGeek]

Generally the transition point where the room takes over on a frequency sweep becomes fairly easy to identify, DSP can be limited to below this.

A lot of people like a downward tilt which often does require DSP at higher frequencies. Then there are roll-offs and shelfs etc. And as Dave said, different windowing etc.

 

If your into this kind of thing, it can be fun to play around with and learn what you prefer with your own speakers in your own room.

[caminperth]

Has anyone used these guys? https://vicoustic.com 

They have a "project request" service which looks interesting and different to the DSP approach.

I'm trying to treat the room physically as much as possible and am not interested in going down the DSP path.

Edited April 20, 2020 by caminperth

[betty boop]

35 minutes ago, caminperth said:

Has anyone used these guys? https://vicoustic.com 

They have a "project request" service which looks interesting and different to the DSP approach.

I'm trying to treat the room physically as much as possible and am not interested in going down the DSP path.

 

yes, @Spizz as reported in thread below, and yes there are definitely folk in this space locally and yes even where you are that provide services likw this. and agree actually correcting the room in this case with treatments does infact correct the room as opposed to DSP and correction curves that correct for... ie rather than actually correcting room in the first place...

 

not to suggest no room for correcting for the room.... something been doing myself for many years...but room treatments and what can do would be the preferential step to do before.

[frednork]

Have been reading carefully what these services linked by the OP offer and it seems to me there is something new here. 

A plugin called Room Shaper that filters unwanted room resonances to create a "perfect" room or in other words, remove unremovable? room resonances. 

 

A detailed review here by mitchco https://audiophilestyle.com/ca/reviews/home-audio-fidelity-room-shaper-review-r855/

 

but to cut to the chase what can it do? sounds very interesting.

 

 

 

If you look at 10 Hz in the top spectrogram and out to just before 400ms, one can see a pretty high level resonance (about -15 dB in level at 400ms with a twinge of green). In the Room Shaper bottom chart, we can see that has been reduced by about -20 dB, so perceptually to ones ears, about one quarter as loud as it once was, at 10 Hz! I know of no other technology that can do this. 

 

[jamesg11]

My understanding from reading thru the above initial 2 online services is that Thierry from HAF developed this ‘room shaper’ & Mitch has included it into his bag of tricks.

[frednork]

4 minutes ago, jamesg11 said:

My understanding from reading thru the above initial 2 online services is that Thierry from HAF developed this ‘room shaper’ & Mitch has included it into his bag of tricks.

I dont think Room Shaper can be implemented without the plugin as it works in real time dynamically and thus cannot be included in a convolution filter. Anyone using it?

[davewantsmoore]

Quote

actually correcting the room in this case with treatments does infact correct the room

 

It's a shame that correcting the in room response got called "room correction" ..... as it really creates a dichotomy with "room treatments".

 

They're not really competitors.... and the idea that (to paraphrase audiophiles generally) someone "wouldn't use EQ, until they'd put up some accoustic panels" ... is quite silly IMVHO.

 

 

[nzlowie]

I'm thinking of trying DRC Designer to create some filters to replace the eq I did in JRiver. I have a question around the sample rate when creating these filters, does it have anything to do with the playback rate of the files being played? Say filters were created at 48khz but the music files are 96khz, any issues?

 

Going to have a play but might get them done commercially. Diy to start....

 

Cheers

[davewantsmoore]

14 hours ago, nzlowie said:

Say filters were created at 48khz but the music files are 96khz, any issues?

No

https://wiki.jriver.com/index.php/Convolution#Automatic_Resampling

 

 

Quote

create some filters to replace the eq I did in JRiver

Whether this is good idea (removing the existing EQ first before running DRC) depends on the purpose of the EQ.

 

Let's say (for example) that your speaker has a big lump or dip in it's response (ie. it's part of the speaker).   Then you should correct that error/isses with the speaker first ... rather than leaving it up to to a "DRC" using data captured from the listening position.

[davewantsmoore]

21 hours ago, frednork said:

Have been reading carefully what these services linked by the OP offer and it seems to me there is something new here.

New....  in so far that it is offered in an audio plug-in that works automagically.

[nzlowie]

5 minutes ago, davewantsmoore said:

Let's say (for example) that your speaker has a big lump or dip in it's response (ie. it's part of the speaker).   Then you should correct that error/isses with the speaker first ... rather than leaving it up to to a "DRC" using data captured from the listening position.

Cheers Dave, I was looking at using the eq to "dial" in some room curve. The response is probably too flat or even a little higher in the 1-2khz range which makes it sound harsh and bright.

Thinking the dedicated filters could perform better than the basic eq options.

 

EQ is the easy way but as you say should work on the speakers first so I might try adding a wide notch filter to drop this range a few db.

[frednork]

38 minutes ago, davewantsmoore said:

New....  in so far that it is offered in an audio plug-in that works automagically.

So, are you saying that this is an "automagic" simple to use  version of something that is done as a matter of course in pro dsp treatment?

 

mitchco seems to think its a new innovation 

 

"For me, Room Shaper is the icing on the cake as I am already using SOTA DSP DRC tech for 3 way linear phase digital crossovers, driver time alignment, room frequency correction and excess phase correction. The ability to control low frequency room resonances and lower mid room characteristics is a new innovation. The ability to control how much is sensed and how much effect is applied (i.e. resonances removed) allows one to fine tune to personal preference. Once I found the sweet spot where the low end sounds focused with no resonances and the lower mids sound crystal clear, no further adjustments were required, no matter what music material was played. So it is set it and forget it."

Edited April 21, 2020 by frednork

[jamesg11]

Interesting - I think I now get the extra significance of this. But I’m not at all sure how Room Shaper is applied in conjunction with say the on-line HAF filters from Thierry.

[frednork]

Its run as an extra thing which has to run in software that supports VST plugins such as Jriver. It seems it can be run with roon controlling Jriver but not 100% on this. 

[davewantsmoore]

On 22/04/2020 at 9:14 AM, frednork said:

something that is done as a matter of course

No.... well, not necessarily.

 

There is no new understanding or technique at play here.  What is new is a plugin with "knobs to twiddle".    However that is a very big deal, as it takes something which is quite complex, and shortens the workflow dramatically.... to the point where machines are doing 99.99% of the otherwsie manual analysis and correction design.

 

Someone might correct one or two major modal resonances like this.... and it might take a lot of work and trial and error.   This plugin lets you correct every resonance, automagically, and then essentailly twiddle a "strength" knob to taste.

 

[davewantsmoore]

7 hours ago, jamesg11 said:

But I’m not at all sure how Room Shaper is applied in conjunction with say the on-line HAF filters from Thierry.

You can apply Room Shaper plgin to your audio independantly of whether or not you apply other filters (like HAF) or not. 

[davewantsmoore]

5 hours ago, frednork said:

Its run as an extra thing which has to run in software that supports VST plugins such as Jriver. It seems it can be run with roon controlling Jriver but not 100% on this. 

Yes, it is just a VST plugin..... can be run in anything which support VST.

[davewantsmoore]

4 minutes ago, davewantsmoore said:

No.... well, not necessarily.

 

There is no new understanding or technique at play here.  What is the new "innovation" is a plugin with "knobs to twiddle".   

 

Predviously, someone might correct one or two major modal resonances .... and it will take a lot of work and trial and error.

 

This plugin lets you correct every mode resonance, automagically, and then essentailly twiddle a "strength" knob to taste..... this shortens the workflow by 99.999%.

 

 

[Mike13]

On 19/04/2020 at 8:56 AM, dan_b said:

The filters Mitch developed 

Who’s Mitch?