bizarre DRC results

coleco · « **on:** 24 Mar 2005, 03:06 am »

I've been playing with DRC for my system but I can't get it to work properly. I'm using edirol ua-25 and a behringer measurment mic (ie, my input is good). I've followed the instuctions exactly as in Denis Sbragions's DRCGuide, DRCGuidev1.0.pdf, but the resulting filter does strange things. The bass seems perfectly out of phase between the channels so when you sit in the listening position there's no bass, but when you move out of the place, the bass comes back. Also it's almost totally notched out the frequencies between 10khz and 20khz.

I've followed the instuctions very closely so don't ask me if I followed them properly, ie, the mic was precisely the same distance from each speaker.. etc.. after following the instructions word for word and finding that it din't work, I got creative and changed the length of the frequency sweeps and edited the drc config files, but the results are always the same.

Any suggestions would be appreciated.

PS

I used drc on my headphones and it seemed to work fine. They sound awesome now.

coleco · « **Reply #1 on:** 24 Mar 2005, 06:12 am »

I played with 'multi-track latency' in cool edit.. in another program it says that latency for my box is 11ms so I entered that and it seemed to help.

I guess the wave form was out of phase slightly.. it doesn't cut the high frequencies now and the bass frequencies aren't out of phase any more.

I recorded a sweep with the DRC filter and the response is basically ruler flat from 30hz up. I mean really really flat.

Subjectively pop music sounds better in that all stridency in the treble is totally gone. Overall the sound is more laidback and not as 'peaky'. Vocals and mids have more 'meat' in general. Maybe it's my imagination, but I think there's a bit of phasyness in the vocals (maybe).

Also the bass is much diminished.. it measures totally flat, but it just doesn't have the punch it did before. I did the frequency response one channel at a time, maybe the channels are still a bit out of phase..
This is harder to get right then I thought it was going to be.

Anyone know how to measure the precise recording latency?

coleco · « **Reply #2 on:** 24 Mar 2005, 07:09 pm »

Okay I tweaked the sweeps.. ie.. tried to elimanate all sources of vibration in the room, elminated as many sources of noise as possible and pushed the volume up to try and get a good S/N ratio.

Results are pretty impressive, I'm still not happy with the bass but maybe I'm just used to a big bass hump. Sound is a bit 'boxy' out of the listening position.. I will tweak the config cause I just don't sit in one place; I need a bigger sweet spot.

Too bad I don't like classical cause it sounds really great. It's *very* convincing.

brj · « **Reply #3 on:** 24 Mar 2005, 07:49 pm »

Unfortuantely, I can't offer you any advice as I'm unfamiliar with your equipment. After a little research into what Edirol offers, however, I'm finding both the UA-25 and your process interesting and informative. Good luck, and please keep us posted on your progress!

geofstro · « **Reply #4 on:** 25 Mar 2005, 07:57 am »

Coleco,

Congratulations for being brave enough to try this, It is tricky to get right; but well worth the effort I believe.

Have you tried this program:

http://pcazeles.perso.cegetel.net/acxo.htm

It makes it far easier to experiment than the method using cooledit and the aurora plug-ins.

You should be able to extract the impulse response file it produces and use it in Foobar.

Patrick is a friend of mine who has an awesome system using compression drivers. You can view them on the web-page. He's also very helpful if you have any questions.

He uses DRC in favour of his Tact RCS system as he feels this totally free software gives better results.

Good luck!

Geoff

coleco · « **Reply #5 on:** 25 Mar 2005, 06:00 pm »

Hi, thank you for the suggestions.

Excuse my vagueness in places I'm still a novice at this.

I discovered the ringing was mostly confined to outside the listening position. It almost disappeared when I moved my head outside that area.

So far I have made the following improvements:

- used 26hz-21khz sweep cause anything below 26hz was resonating weirdly through my c5s.

- tried to eliminate all noise in the environment during the sweeps and pictures and etc that were vibrating off the wall.

- used 'soft.drc' to increase the 'sweet spot'

- changed all the *startfreq references to 26hz

- changed PSPointsFile to bk-2-sub.txt to eliminate any subsonic distortion.

- after the first channel correction entered the impulse center into the config as a reference to the other channel

Phasiness in the vocals is much diminished now. Still working on it.. As so as I understand what these config files do better, I will combine some of the changes in optimized.drc into soft.drc.

thayerg · « **Reply #6 on:** 25 Mar 2005, 11:24 pm »

I second the congradulations for attempting this. I read the instructions about DRC but then chickened out. I like the looks of the software geofstro mentioned.

Does anyone know whether the rat shack spl meter mic is good enough to attempt DRC with? I know there are published compensation curves for the bass but i'm thinking that the bass is really where DRC needs to be most precise.

coleco · « **Reply #7 on:** 27 Mar 2005, 03:19 am »

The rat shack spl meter probably isn't good enough, it has a rca jack out, but the noise floor of the signal is way too high. Also the response of the mic is only calibrated to 10khz (at least the one I bought).

I can't use ACXO cause I have 4 sound devices and I just can't get it to work.

I think my problem is with Adobe Audition. It's very buggy. I just made some really nice 96kbps sweeps but the generated impulses are screwed up. I'm going to use the deconvolver supplied with drc to see what happens.

coleco · « **Reply #8 on:** 27 Mar 2005, 07:11 am »

I eliminated the ringing by reducing the size of the phase band windows. I also tweaked a couple other settings.

Imaging is still confused probably because the main impulse spikes of one of the channels is between the samples and the filter comes out misaligned.

coleco · « **Reply #9 on:** 29 Mar 2005, 12:53 am »

I re-recorded the sweeps in 48k.. They turned out nice and the impulses have better first peaks.. so I fed them into drc and it's sounds pretty good. Day and night compared to no drc. The focus in the soundstage isn't perfect but I can live with it.

Remember.. digital 'room' correction is a bit of a misnomer because it will correct problems with your speakers as well.. It elminates harmonic distortion and even time aligns the waveforms (because usually the drivers cones aren't time aligned'

My combo, energy c5s and a panny xr50, has retained it's signature revealing qualities but it now sounds compeletely tonally balanced, the highs are laid back and the mids are nice and full.

I have a feeling the ringing in the default config was caused by the program trying to overcorrect for my crappy room and room placement (which currently I can't do anything about).. basically setting the config to be less agressive stopped it.

I'm going to play around more with the config and see what happens...

coleco · « **Reply #10 on:** 1 Apr 2005, 08:23 am »

Made some more changes:

- increased *FilterLen in a couple places
- increased *MultExponent in a couple places
- included PSMicPointsFile = ecm8000.txt and PSMicMagType = D to correct for the freq and phase response of the measurment mic.

The combination really opened up the soundstage, which was lacking before. Listening to Reflective Frequencies by Solar Fields, which is pretty much the gold standard for production quality for me, the speakers really disappeared and the holographic quality of the album made go 'oh wow' a couple times-- which hadn't happened up until then. The only drawback of increasing those values is that the filters take longer to compute (who cares).

MPWindowExponent seems to be the variable that when reduced will reduce the ringing. According to the readme, increasing the value increases the correction in the midrange. 0.8 works for me, I'm going to try tweaking that one more.

One thing I'd like to do is to increase the number of taps in the final filter, I'm not sure how to do that though.

I'm really impressed with the sound that's coming out considering the room placement.

geofstro · « **Reply #11 on:** 2 Apr 2005, 08:47 am »

You're doing a great job!

Do you think you'd be able to write a short guide on these tweaks and post it here?

I simply followed the same DRC guide as you did to the letter. An additional guide to the various tweaks that are possible and the benefits of playing with those values would be very useful.

Thanks again

geoff

coleco · « **Reply #12 on:** 4 Apr 2005, 07:53 pm »

Yeah when I have a better understanding of how this works I will.. a good place to start however is the readme.. If put together a guide at this point I would be mostly talking aut of my ass