Using convolver universally in HTPC

**Fujak** · **05-26-09, 12:02 PM**

Thanks a lot for your help. After all I think also, that the most important thing is to build a correct impulse response. So I will try out REW to build the impulse response and will give a feedback about the result.

**Fujak** · **05-27-09, 03:37 PM**

Antani, now I am back. I have tried out REW. I can make the measure. After this I go to menu "file->export->export->Impulse Response as WAV..."
The exported file ImpulseResponse.wav has to be the BCInFile in DRC, right? The result (the filter) however looks not o.k.
Did I make something wrong? I guess the BCInFile has to be in PCM (raw) format. But in RWE I can export the Impulse Response only in WAV-Format. Do I have to convert the exported file first, an if yes how ist the sox-script for this?

**antani** · **05-28-09, 03:10 PM**

DRC accepts only pcm 32 bit floating.
The command in SOX is

sox.exe impulseresponse.wav -t raw -r 44100 -c 1 -f -4 impulseresponse.pcm

**cea** · **06-01-09, 03:14 PM**

Hi.
I have a config file for 44100 and another for 48000, how can I make a config file for switch automatically between filters?.
Here http://convolver.sourceforge.net/config.html says that can switch automatically but I don't know how.
Thanks.

**Fujak** · **06-03-09, 01:07 AM**

Quote:

cea wrote:

Hi.
I have a config file for 44100 and another for 48000, how can I make a config file for switch automatically between filters?.
Here http://convolver.sourceforge.net/config.html says that can switch automatically but I don't know how.
Thanks.

I even don't know, but I guess, they mean the button at the top, where you can select various correction files in order to load them into the convolver.

**antani** · **06-03-09, 11:36 AM**

I don't use Convolver so I'm not really sure. However from what I understand, you should create multiple settings in the config.txt file, simply putting them one after the other, in different rows. Convolver should choose automatically the appropriate settings from the sample rate of the file being played.

From the Convolver help:

So a config file can also comprise of a list of

- filter specification config file names of the type described above; and
- WAV impulse response filenames

one per line. The filter used will then be the first to match the current sound source (in terms of number of input and output channels and sample rate). This allows you to play both stereo and 5.1 sources, say, without having to change the config file.

**cea** · **06-03-09, 01:46 PM**

Hi, thanks.

Quote:

I even don't know, but I guess, they mean the button at the top, where you can select various correction files in order to load them into the convolver.

you mean "Get config or IR" button? That button loads only one file, I can not load more than one.

Quote:

I don't use Convolver so I'm not really sure. However from what I understand, you should create multiple settings in the config.txt file, simply putting them one after the other, in different rows. Convolver should choose automatically the appropriate settings from the sample rate of the file being played.

From the Convolver help:

So a config file can also comprise of a list of

- filter specification config file names of the type described above; and
- WAV impulse response filenames

one per line. The filter used will then be the first to match the current sound source (in terms of number of input and output channels and sample rate). This allows you to play both stereo and 5.1 sources, say, without having to change the config file.

well, I try to make one config file with both but I cann't find the way, one after the other or in different rows... nothing.
Thanks.

**antani** · **06-04-09, 05:10 AM**

The syntax of Convolver isn't really user friendly

. I also remember that some people had problems with the format of the config file (that should be Unicode or ANSI, I'm not sure).

**danik** · **06-17-09, 02:50 PM**

Quote:

antani wrote:

I don't use Convolver...

Hello Antani,
what to use instead of Convolver?

Milan are also me, we can meet?

ita2eng by google, sorry

**pakru** · **07-08-09, 03:27 PM**

Quote:

antani wrote:

It's not bad, but I think you can get a better result. In my environment I have a flat frequency response +- 1 dB, with a smoothing of 1/3 oct. Try the other .drc presets. But in particular pay attention not to clip the measuring sweep (I'm using 8 sweeps of 1 M).

DRC can create minimum phase, linear phase and mixed phase FIR filters. All the .drc presets are mixed phase. Mixed phase means the filter is correcting the minimum phase component and a part of the excess phase component. It's a sort of compromise between minimum and linear phase, with a short preringing.

If you try to average the sweeps, you will only get a bad filter. DRC is not Audyssey (luckily

).

Hello antani, how do you use 8 sweeps of 1M? Do you average them? Thanks.

**antani** · **07-09-09, 09:04 AM**

Quote:

pakru wrote:

...Do you average them?...

No need for that. Just select 8 sweeps in the Measure window. By the way I think there is a little bug in the sw, because I often get weird results. I always check the coherence with a single short sweep, and the scope window as well (sometimes clipping is not signaled).

**brucek** · **07-09-09, 09:54 AM**

Quote:

By the way I think there is a little bug in the sw, because I often get weird results.

We do get that comment now and then, but in the end it usually ends up being a problem with the computer running REW.

This post (to the end) is a perfect example, where in the end, the problem is computer related.

brucek

**antani** · **07-09-09, 11:21 AM**

It is definitively possible, I didn't test very much. However I got problems only with multiple sweeps. I guess the multiple sweeps option is seldom used, but it is very useful when maximum resolution is needed, like for DRC.

**brucek** · **07-09-09, 12:25 PM**

Quote:

I guess the multiple sweeps option is seldom used

I always use at least two. This reduces random noise that is picked up in a single sweep.

brucek

**pakru** · **07-09-09, 01:40 PM**

Quote:

antani wrote:

No need for that. Just select 8 sweeps in the Measure window. By the way I think there is a little bug in the sw, because I often get weird results. I always check the coherence with a single short sweep, and the scope window as well (sometimes clipping is not signaled).

Thanks for pointing this out. I did get very weird results with multiple sweeps (wierd comb like transfer function) and I was thinking there is something wrong with my setup. Can you help me understand how you check for coherence?

Also, I am using console with the optimized.drc coefficients loaded in Voxengo PristineSpace. When I test the room correction filter by sweeping REW through PristineSpace in console, REW is complaining that the impulse response peak is not where it is supposed to be. Looking at the PristineSpace window, I find that the filter center tap is at the middle of the window (latency of N/2 samples). Is this the reason REW is saying this?

Thanks.

Quote:

brucek wrote:

I always use at least two. This reduces random noise that is picked up in a single sweep.

brucek

brucek, Even with two sweeps, I am getting a wierd comb like transfer function with the first null at 1.75KHz. I can try to post the plots after I get back home.

**brucek** · **07-09-09, 01:45 PM**

Quote:

I am getting a weird comb like transfer function with the first null at 1.75KHz

With that and REW complaining about the impulse response, it sounds like your setup levels are incorrect.

brucek

**pakru** · **07-09-09, 01:49 PM**

However, with a single sweep, I dont see any combing artifacts...

**brucek** · **07-09-09, 04:21 PM**

Quote:

However, with a single sweep, I don't see any combing artifacts...

Then you may be suffering from a lack of processor speed or memory. The longer the sweep, the more samples that must be stored, and the number of sweeps simply adds to the problem because it all has to be stored in RAM.

brucek

**pakru** · **07-10-09, 02:27 AM**

antani, brucek:

Thanks to antani's tip of looking at the scope window, whenever I use a multiple sweep option in the measurement, I am seeing wierd looking input samples, only in the second sweep (figure below). I dont see this behavior in the first sweep. Whenever this happens, the transfer function has the combing artifact.

Is this my setup issue or is it something else?

Thanks.

**brucek** · **07-10-09, 08:32 AM**

The input is the dark red line and the output samples are the purple.

Looking at the time axis, I would presume you're examining a fairly high frequency. You can expand the axis and look at each sample. You'll probably find that the 'sinusoidal' wave is distorted at this higher frequency.

I know on my inexpensive soundcard, that the DAC probably isn't the greatest, and regardless of whether I use more than one sweep, I'll have a fairly poor looking output signal at the upper frequencies. Certainly at 20K, with a 44KHz sample rate, there's only a couple samples per period and it looks pretty bad.

I'm not sure why you would only see the output problems on multiple sweeps though. I'll ask JohnM what he thinks..

brucek

**antani** · **07-10-09, 08:46 AM**

This comb effect is pretty strange. I normally check the multiple sweeps result with a single short sweep. If the frequency response is very similar, the measurement is supposed to be ok. In the scope window I just check the clipping.

I often get problems with 8 sweeps, rarely with 4 sweeps, so it is possible that the issue is lack of memory as Brucek was suggesting.

The warning "response peak is not where it is supposed to be" is normal if you use a linear phase FIR filter. In fact the linear phase filter adds a delay of 0,7 s, but this shouldn’t cause any problem. BTW you don't really need to use a linear phase FIR, unless you are using a brickwall filter.

**JohnM** · **07-12-09, 07:51 AM**

Quote:

brucek wrote:

I know on my inexpensive soundcard, that the DAC probably isn't the greatest, and regardless of whether I use more than one sweep, I'll have a fairly poor looking output signal at the upper frequencies. Certainly at 20K, with a 44KHz sample rate, there's only a couple samples per period and it looks pretty bad.

The sweep signal REW shows is what is fed to the soundcard rather than what the DAC actually produces, which REW has no access to. As the comment on the scope page says, "These traces use linear interpolation between values, this gives them an aliased appearance at high frequencies that is not a true representation". The scope trace on the screen shot shows normal appearance. If the PC is prone to struggling to keep up with soundcard I/O, which can be the case depending on what other processes are running, then the longer the sweep sequence the more likely there is to be a glitch in the I/O. 8 sweeps at 1M requires more than 3 minutes of clean I/O, so if there is a problem that will certainly show it up.

**pakru** · **07-13-09, 11:33 PM**

JohnM, Thanks for the explanation. I will try the sweeps from a different PC and see if this persists. If the problem persists, I will try to up the RAM on the regular PC.

Thanks again!

**pakru** · **07-13-09, 11:34 PM**

Quote:

antani wrote:

The warning "response peak is not where it is supposed to be" is normal if you use a linear phase FIR filter. In fact the linear phase filter adds a delay of 0,7 s, but this shouldn’t cause any problem. BTW you don't really need to use a linear phase FIR, unless you are using a brickwall filter.

Thanks!

**nbb** · **07-16-09, 03:31 PM**

Thx for a great thread with spot on information for my use. Im struggeling a bit with console, could anyone be so kind as to how they set up the routing for the various modules? I assume one just add console as an external filter in Media Player Classic to have the sound pass through the plugins hosted by console. Im trying to have a simple 2-channel signal passed through convolver and the Kelly Industries Bass Manager for a start.

Advertisement