Quantcast

acourate: an audio toolbox

mattkhan

Distinguished Member
Does anyone use any PC based DRC solutions like acourate, audiolense or similar?

I am currently building a new server which deals with audio and video streaming so considering going down this route for a more powerful & customisable eq solution over that found in a processor hence interested in any feedback anyone has.
 

mattkhan

Distinguished Member
I have taken up the "try acourate at home" offer as described on their site

The comparison between audyssey multeq xt (on my Marantz AV 7005) and the acourate filtered version is quite interesting, the difference is both subtle and marked. One of the tracks I gave him was "Spirit in the Night", a track which is dominated by a sax and organ/piano (at least in the early part). On my system, the multeq rendition is really dominated by the sax whereas the acourate version feels more like the organ is leading with the sax providing highlights, i.e. it's a noticebly warmer rendition. The latter is certainly more enjoyable to listen to though I have no idea which is more technically accurate.

It would be interesting to know what the filter looks like as the frequency response in my LP certainly "enjoys" the usual floor bounce depression in the 250-350Hz range. If this has been corrected then it may well correspond to the change I have heard.
 

mattkhan

Distinguished Member
I received new versions of the files yesterday which incorporate my mic cal file (I have an individually calibrated mic) hence a more accurate filter can be created. Quite remarkable difference, I could wheel out all the cliches at this point (a veil being removed etc etc) but it really was an immediately obvious difference. The whole sound stage just hung together that bit more coherently and each individual instruments that bit more realistic and well defined. The slight wandering in the sound stage, when the L and R responses vary in amplitude across the frequency range, really disappeared. Highly recommended.
 

KelvinS1965

Distinguished Member
I've only recently got myself an Audyssey equiped AVR so I wasn't aware (though not surprised) there is a PC based solution. The only issue I have is that I lost patience with my previous HTPC as any updates seemed to upset it and I must have re installed it (or rather re 'cloned' my C drive) many many times.

Are you outputting from the PC in analogue or via digital means?
 

mattkhan

Distinguished Member
There are a few different solutions for a purely PC based solution; Dirac live, audiolense, accurate are just 3 of them. My current setup is stereo audio via optical and multichannel via HDMI. The latter is really unused ATM, I am moving to a jriver based setup once I have built a new server though (at which point that will take over from my ps3 for bd duties).

This article gives an overview of acourate - Computer Audiophile - Acourate Digital Room and Loudspeaker Correction Software Walkthrough
 

xsnv

Member
I am very interested in this as well. Was going to put up a thread up in the speakers forum (larger readership) to discuss.

I play most of my media via my Mac mini (and an external drive). The Anthem does a good job but as I have the processing power on tap anyway It'll cost less incrementally.

What informed your choice Matt? Which suites are considered the best in the market?

Regards.
 

mattkhan

Distinguished Member
My requirements were that the tool;
  • can be used under windows or linux
  • can be used for multichannel as well as stereo
  • can be used to apply a digital crossover
  • allows user to fully control the target curve (from sub signal shaping to full range drc)
  • has at least some sort of guided mode (or recommended default settings)
  • is well supported or at least has some sort of community around it (as I fully expect the learning curve to be a cliff face)
The latter requirements were added after I starting looking into detail at brutefir which is just the purest wtf even for someone who is happy at the linux command line. The windows + linux requirement came in after I realised that my life is not long enough to consider going back to a linux htpc & that jriver really looks like the only game in town when it comes to a really audiophile media centre.

I then spent some time trawling the computeraudiophile forums and boiled the choice down to audiolense, acourate, dirac and drc-fir. Dirac fails the cross platform requirement as it is windows only and relies on its player being in the chain. drc-fir fails on the guided mode requirement so I was down to acourate and audiolense. Uli was helpful & responsive over email, has a better demo offering (try your own music after he convolves it) & I v v much liked the results hence choice was made without trying audiolense. As far as I can tell there are v happy users of both audiolense and acourate so I doubt you'd go too far wrong either way, I believe the guys behind them are at the forefront of the field at any rate.

Hopefully I'll have my new HTPC built in the next few days so I can then actually start playing with it :)
 

mattkhan

Distinguished Member
Interesting example of the flexibility of acourate here - Computer Audiophile - Advanced Acourate Digital XO Time Alignment Driver Linearization Walkthrough

I have been doing an exceptionally crude form of this using rew to time align sub to mains (ie take a few measurements, try and match up impulses and phase as best you can) but get frustrated at being left at the mercy of a black box (the processor bass management) whereas this should (I hope) let me take control of that and make it exactly right.

The linked article is going to need to be read many times mind you, quite complex to say the least! I imagine some years might pass before I get there :confused:
 
Last edited:

sergiup

Distinguished Member
Thanks Matt, going to follow this with interest!
 

mattkhan

Distinguished Member
I think what I'm aiming to achieve in acourate is 3 things
  1. extending the low end of a sealed sub
  2. bass management
  3. full range room correction
On the 1st point, there is a suggested approach for a linear phase linkwitz transform based on a near field measurement of the sub, manually correcting that (using acourate's target curve controls) to get the response you want and then passing that through 3-4 more steps using some of the tools in acourate to deliver a linear phase transform (hence no group delay impact). This sounds fairly straightforward to do (famous last words) and much less crude than my current "whack a LS12 filter on from ~25Hz in the inuke" :)

I don't currently remotely understand the mechanics of putting 2 & 3 together, suffice to say it sounds complicated :confused: but hopefully the mists will clear in the next few days.
 

mattkhan

Distinguished Member
A light bulb moment tonight as I now see that bass management is performed by calculating separate mono correction files for each channel (1 low pass, 1 high pass) and then routing them to the output channel using a convolver config file (Config file
Interestingly this opens up possibilities like a dual sub setup that could be used as a full range pair for stereo duty (just by swapping config files).

Now I just need to work out the measurement method. I think it means measuring a channel at a time, applying an xo then correcting but not 100% sure how to get the signals routed ccorrectly over hdmi.
 

mattkhan

Distinguished Member
I am starting to understand the mechanics of taking measurements now, albeit on paper only so far. Tomorrow night will be my 1st try at taking measurements (expletives expected).

I am now understanding why anyone doing this kind of work uses a pro audio interface, i.e. to provide the ability to arbitrarily route channels in software alone. There is an alternative approach for those poor souls connecting via hdmi which involves swapping cables between sweeps manually. Therefore I've been researching reasonably priced pro audio interfaces that have sufficient physical outputs, so far the focusrite saffire pro 24 looks like the best bet as it's the only one I have found under £200. It's not as tidy as the focusrite scarlett 18i20 as it is not rack mounted but it is half the price so that's nice.

It appears extremely easy to spend a lot of money here but I think it will be completely pointless unless I get rid of the processor completely or replace it with one that has a pure MC analogue passthrough.

It's a good job I went into this thinking it was going to cost me more money than I initially budgeted for :zonked:
 
Last edited:

mattkhan

Distinguished Member
The saffire pro 24 turned up from thomann.de within 2 days which is nice and quick. This was the cheapest place I could find it new.

Annoyingly the manual online says it comes with a FireWire 400 cable (6 pin at each end) but it actually comes with a 6 pin to 9 pin cable (FireWire 800) hence a trip out to buy an overpriced cable as I obviously couldn't wait a day or so to save £6 by ordering from Amazon :rolleyes:

Anyway I am now equipped to take measurements without messing around with cables.

I think I could also use this to route my TiVo through the jriver audio engine too. I am thinking optical from TiVo to saffire to the computer through the jriver engine (and hence all the DSP) then back down to the saffire and onto the mc analogue in on my processor. Lip sync could be the problem here as a linear phase acourate convolution can incur a second or more latency hit so I imagine I would need to use a cut down convolution (or just use jriver peq alone).

One day soon I will actually listen to something :eek:
 

mattkhan

Distinguished Member
OK I am now equipped with sound from my HTPC, the measurement process for 2.1 is

  • generate a 2 way crossover (2nd order neville-thiele at 90Hz in my case)
  • save this as a multiway wav which is a single WAV file with 4 channels (Lsub, Rsub, L, R) that will play back through the logsweep recorder as if it is a L and then R sweep
  • configure mixer to route those 4 channels to the right physical outputs
  • sweep and then deal with the room correction

The 3rd step had me mildly confused at first as I've never used mixing software before but it's quite simple once I got the terminology. In saffire mixcontrol you can create mixes from inputs and then route those mixes to other mixes or to outputs. I have the line outputs configured as per hdmi for simplicity (1 = L, 2 = R, 3 = C, 4 = SW, 5 = SL, 6 = SR) so this means I want to route

1 -> 4
2 -> 4
3 -> 1
4 -> 2

to do this in mixcontrol you

DAW1 -> mix1
DAW2 -> mix1
DAW3 -> Line Output 1
DAW4 -> Line Output 2
mix1 -> Line Output 4

(where DAW refers to a channel coming down from the computer to the audio device)
 

mattkhan

Distinguished Member
Miracles do happen, I finally have a working 2.1 setup in jriver with the xo implemented by acourate.

Early days but a definite improvement, greater clarity and a certain "airiness" that wasn't there before which is v pleasing on the ear.

The routine for generating the filters is surprisingly straightforward, just a few macros to run in acourate that step you through the process. You then write a convolver cfg file to load into jriver and away you go. It could certainly be more automated but it is not complicated in itself.
 

mattkhan

Distinguished Member
I have spent more time with this and am starting to get somewhere though no actual new setup to report yet. It seems the approach is to linearise the mains based on near field measurements, apply appropriate timing adjustments to align each main with the sub individually and then apply room correction on top of that.

So far so sensible and doesn't seem *that* hard to implement. The one thing that is certain though is I need to take a day off in order to actually get the work done!! It is not a quick process that's for sure.
 

Jameskatie

Distinguished Member
It can soon zap a day away messing with settings and taking measurements, I forgot to have dinner when setting mine up started at 9am next thing I know it were 6pm and I was sat on the floor with my earphones on running sweeps lol
 

BobbyMac

Banned
It can soon zap a day away messing with settings and taking measurements, I forgot to have dinner when setting mine up started at 9am next thing I know it were 6pm and I was sat on the floor with my earphones on running sweeps lol
Lies, going by your avatar you've never skipped a meal/carbs in your life :p
 

Jameskatie

Distinguished Member
Haha, couple of years ago I wouldn't have done, I've lost a lot of muscle since that pick, the joys of going self employed :)
 

mattkhan

Distinguished Member
I actually made real progress today :clap:

I worked my way through mitchco's approach to linearising & time aligning each way in the system and got the results loaded up in jriver. I've stuck to a "simple" 2.1 setup so far as it's quite time consuming, it took me ~3.5hrs end to end for linearising 3 speakers & then aligning/correcting them as a 2.1 setup, and I want to be sure the 2.1 setup is optimal before I spend what likely be most of a day on a 5.1 setup. The other reason is that my system is tilting towards music being the majority use & eq is more obviously good or bad in a stereo setup.

Anyway the good news is it is a marked improvement on the basic correction (i.e. stick a mic down, take some measurements, correct), v enjoyable to listen to with a satisfying clarity and balance. I don't think it's quite right though on a few levels;

* the gain structure seems wrong now, I have to crank the levels up quite a bit which raises the noise floor somewhat (I've noticed that every filter I create produces a different gain so perhaps I am missing some step in the workflow or have jriver configured incorrectly)
* I am somewhat convinced the soundstage is wandering at certain frequencies, my wife doesn't agree at all so maybe I'm just providing a rationalisation for "can you take the kids out for the day so I can get my mic out and stare at graphs" :zonked:. I'm not convinced mind you, if it does exist it must be at a quite specific and narrow frequency range.

I think a pause to listen for a few days is in order now.
 

mattkhan

Distinguished Member
Further reading since the weekend suggests that my attempt to linearise the speakers was almost certainly flawed as I was measuring too close to the speaker. The open question is whether linearising a passive multi driver speaker when you can't measure in freefield is even a worthwhile exercise but I will come back to that another day.

I also had the opportunity to have Uli (author of the software) walk me through generating a correction today which was enlightening. This included getting the latest build of the software which provides a measurement based mechanism for positioning the mic so it is equidistant from the 2 speakers for the perfect stereo image. This will also then tell me the offset between speakers so I can adjust that electronically (at sample rate precision, so I think that means 3.4mm at 96kHz :))
 

mattkhan

Distinguished Member
I have to say that initial impressions of the correction Uli produced are simply :clap: It's difficult to describe the sound in any sane terms, it is just immediately foot tappingly (I am wearing slippers as I type :smoke:) good. A few moments have reached into hairs on end oh my days territory. The mind boggles at what one could do with a truly dedicated room.

Given this is an EQ change only (albeit one that requires you to use a PC as your only source and invest a shed load of time learning how to use it), cheap at half the price is the phrase that springs to mind.
 

mattkhan

Distinguished Member
The clarity and transparency is pretty remarkable, to give an example every individual strand of percussion is now revealed such that listening to music is actually quite hard to do passively. Your ears just keep pricking up as in "did I just hear that??".

Next step is multi channel. I think I have a good handle on the approach now, just need a clear half day or so to measure up. Once I get that done then it is time to start on the 3d LUT :facepalm:

This is all a good incentive not to upgrade again. I would need to factor in a weeks holiday afterwards!
 

sergiup

Distinguished Member
Matt, I just wanted to say thank you for keeping us updated! As far as I'm concerned, EQ/room correction on a PC is long overdue, especially considering that hardware audio processing is now back in Windows for example and makers such as AMD are open to implementing DSPs in hardware.
 

mattkhan

Distinguished Member
Matt, I just wanted to say thank you for keeping us updated! As far as I'm concerned, EQ/room correction on a PC is long overdue, especially considering that hardware audio processing is now back in Windows for example and makers such as AMD are open to implementing DSPs in hardware.
Modern hardware is so powerful that bespoke hardware is unnecessary, my machine is barely taxed during stereo playback for example. CPU utilisation is a few percent, convolution filter is running at 18x real-time and this is with everything upsampled to 96kHz with 64bit resolution in the processing chain. This is miles ahead of consumer hardware. I am using madvr with all settings close to max and this causesa £150 GPU to notice but it is still not stressed at all.
 

Similar threads

Trending threads

Latest News

AVForums Podcast: 31st May 2020
  • By Phil Hinton
  • Published
MQA expands global partnerships for high end experience
  • By Andy Bassett
  • Published
Samsung 2020 TV app lineup upgraded
  • By Andy Bassett
  • Published
Film festivals move online: We Are One starts 29th May
  • By Andy Bassett
  • Published
Bafta 2020 TV Awards to proceed behind closed doors
  • By Andy Bassett
  • Published
Top Bottom