Amarra Symphony with Dirac Live Impulse Response Correction

RBFC · Dec 19, 2013

Well, here's an interesting set of graphs to illustrate the magnitude of improvement possible with this correction software. Dirac is built-in to the Amarra Symphony playback software. Since I already owned Amarra, I paid the small fee and activated the "IRC" function. You may then take microphone measurements as directed by the software, then generate the filter which corrects the response at the listening position.

Here's the frequency response spectrum of my room (awful) without any correction. I was using some simple parametric EQ to smooth out the lower regions, but nothing close to the results achieved with the Dirac.

The "target" frequency response curve is shown in orange.

Lee

RBFC · Dec 19, 2013

Here's the frequency response after correction to the desired target curve:

Lee

RBFC · Dec 19, 2013

Here is the impulse response before correction:

Lee

RBFC · Dec 19, 2013

Here's the impulse response after correction:

Lee

RBFC · Dec 19, 2013

I cannot overemphasize what a tremendous difference this software makes to the listening experience. While I know that my room is far from ideal, I've had many audiophile friends that thought things sounded very good (remember I was using some primitive EQ, so the shown graphs are not quite representative of the actual listening conditions).

The changes brought about by Dirac:

1. Imaging is much tighter and more believable. A voice sounds much clearer, and instruments remain the same size as they play up and down their respective registers.

2. The "see-through" clarity is a gargantuan step ahead. It's almost eerie as you listen to recordings you thought you knew. This is another case of wanting to revisit your entire collection.

3. The "bloat" and overhang is gone. Period.

Actually, it was hard to listen to the system at first, since I had become so accustomed to "hearing through" the previous distortions that were present. It didn't take long, though, to realize that something special was going on, and that many of the proverbial "veils" had been lifted. You may switch the correction on and off instantaneously if you want to do comparisons. I didn't last long listening to things as they were for very long!

Many of you may say that your room/response is far better than that in the pictures. Fair enough, but I'd estimate the correction as being the equivalent of 10's of thousands of dollars in hardware upgrades. Money well spent.

Special WBF thanks to dallasjustice, whose generous assistance got me through this without inducing a cerebral aneurysm.

Lee

amirm · Dec 19, 2013

No audio system is working right without EQ. It simply cannot. Even simple things like level and timing matching with auto eq performs can have a drastic difference in imaging.

That said, the correction in my opinion may have gone too far. To get such a flat response, it would have had to pull the overall level down, and then attempt to amplify them to get the gain back. If so, you will be pushing your amplification and speaker system hard. Keep an eye on the temperature of the amp and make sure it is not overdriven.

But congrats in moving in this direction. It is startling how much improvement one can get here. It truly is night and day

.

RBFC · Dec 19, 2013

Thanks, Amir. My Krell EV403 doesn't work very hard to play very loud, so I don't think it's a problem. The preamp volume control does not require much more gain to achieve listening levels, if any at all. The difference is amazing. I thought it would be educational to show how bad a "good-sounding" system can measure, and what magnitude of improvement can be gained. I did not see such crazy response using the app on my iPhone called Spectrum Analyzer, so a calibrated quality microphone is truly necessary to get this right. The biggest changes are in the response of my twin JL F113 subs, which each use 2500 watt amps. I have heard them play much louder than I ever listen to them during their calibration and demo modes, so I think we're OK there, too.

Note the dB scale on the left to see where the levels have been changed. It appears that ~ 55Hz has been boosted by 10dB, which lies in the subs' range. At any rate, the software has built-in boost limiting, to prevent overloading your amplifiers. It looks like the 10dB boost is the limit, since it left that dip at 55Hz down about 5dB from the target curve.

Lee

edorr · Dec 19, 2013

amirm said:
That said, the correction in my opinion may have gone too far. To get such a flat response, it would have had to pull the overall level down, and then attempt to amplify them to get the gain back. If so, you will be pushing your amplification and speaker system hard. Keep an eye on the temperature of the amp and make sure it is not overdriven.

Keep in mind though the "after correction" curve is not an actual measurement, but a theoretical curve. If you measured the response after correction independently using something like REW, you actual in room response would not look nearly as flat.

RBFC · Dec 19, 2013

edorr said:
Keep in mind though the "after correction" curve is not an actual measurement, but a theoretical curve. If you measured the response after correction independently using something like REW, you actual in room response would not look nearly as flat.

Yep. But it sure sounds a lot better. I'd be very interested to see the before-correction curves of some of the super systems we have here on WBF!

Lee

dallasjustice · Dec 19, 2013

It's really not possible to do this. I've thought about how it could be done in REW and I don't have a good way to do it. You can't simply take a single REW measurement through the DIRAC filter and call it a day. You would need to take 9 measurements and then know the proprietary method that DIRAC uses to combine all of those measurements. It's one of the those exercises that is pointless really. Having said that, I believe wgscott averaged all the measurements and they came out very close to the predicted measurements. You can read about that here:
http://www.computeraudiophile.com/f23-dsp-room-correction-and-multi-channel-audio/experiment-17437/

edorr said:
Keep in mind though the "after correction" curve is not an actual measurement, but a theoretical curve. If you measured the response after correction independently using something like REW, you actual in room response would not look nearly as flat.

edorr · Dec 19, 2013

RBFC said:
Yep. But it sure sounds a lot better. I'd be very interested to see the before-correction curves of some of the super systems we have here on WBF!

Lee

Your measured pre-correction in room response is not bad at all +/- 10db. Not unlike my room before I put in massive corner traps, which improved things quite a bit.

I will be getting my server with a copy of Dirac live back soon myself. Not sure yet if I'll end up using it for 2 channel.

edorr · Dec 19, 2013

dallasjustice said:
It's really not possible to do this. I've thought about how it could be done in REW and I don't have a good way to do it. You can't simply take a single REW measurement through the DIRAC filter and call it a day. You would need to take 9 measurements and then know the proprietary method that DIRAC uses to combine all of those measurements. It's one of the those exercises that is pointless really. Having said that, I believe wgscott averaged all the measurements and they came out very close to the predicted measurements. You can read about that here:
http://www.computeraudiophile.com/f23-dsp-room-correction-and-multi-channel-audio/experiment-17437/

May be more difficult than I thought. I thought you can run a signal generated by REW through the filters and measure actual in room response. May be I'm wrong (probably am).

dallasjustice · Dec 19, 2013

amirm said:
No audio system is working right without EQ.

+1

dallasjustice · Dec 19, 2013

edorr said:
May be more difficult than I thought. I thought you can run a signal generated by REW through the filters and measure actual in room response. May be I'm wrong (probably am).

Yes, you can do that. But, that would just be one measurement. One of the big problems with DRC is the single measurement problem with an omnimic. It's really not possible to get an accurate measurement (what you actually hear) using only a single omnimic measurement from seated position. I think that's one of the places where DIRAC distinguishes itself from many other DRC softwares.

edorr · Dec 19, 2013

amirm said:
No audio system is working right without EQ. It simply cannot. Even simple things like level and timing matching with auto eq performs can have a drastic difference in imaging.

Which does not mean it is always a net improvement. There is typically a trade off. I spoke with some guys (blue smoke) that set up a TAD room with the TAD chief designer (name escaped me). They ran Dirac on their server, got great improvement in "room interactions", but some loss in transparency, and decided against using it.

dallasjustice · Dec 19, 2013

I agree there can be a trade-off. However, there's no trade-off in my system. Everything is better using DIRAC.

Your server has been out of commission for a while now, right? Are you still using Trinnov too? I've heard good things about Trinnov also. Nyal said he might be getting one soon. I would love to hear his thoughts on it as well. I have heard the DEQX at a show and was very impressed with what it could do with Radioshack speakers.

I don't think there's any single solution that works for everyone. But, everyone will benefit from using EQ, if done right.

edorr said:
Which does not mean it is always a net improvement. There is typically a trade off. I spoke with some guys (blue smoke) that set up a TAD room with the TAD chief designer (name escaped me). They ran Dirac on their server, got great improvement in "room interactions", but some loss in transparency, and decided against using it.

edorr · Dec 19, 2013

dallasjustice said:
Your server has been out of commission for a while now, right? Are you still using Trinnov too? I've heard good things about Trinnov also. Nyal said he might be getting one soon. I would love to hear his thoughts on it as well. I have heard the DEQX at a show and was very impressed with what it could do with Radioshack speakers.

It has been out for repair for a while. First diagnosed with defective motherboard, now harddrive also turns out to be defective. Should have it back in a week or two I hope.

dallasjustice said:
Are you still using Trinnov too? I've heard good things about Trinnov also. Nyal said he might be getting one soon. I would love to hear his thoughts on it as well. I have heard the DEQX at a show and was very impressed with what it could do with Radioshack speakers.

I only use the Trinnov on my MCH setup, because it has only AES/EBU. USB into my DAC is better than AES/EBU, offsetting the benefits of using Trinnov in the 2 channel signal path. My plan is to eventually just run Dirac on my server and run MCH Lynx straight into 3 x 2 channel DACs, and sell the Trinnov.

dallasjustice · Dec 20, 2013

More shots

I thought I would post some shots of my measurements. These are before/after measurements. I also posted some shots of the new version of DIRAC. I am showing the DSP gain function.

Amir brought up a good point about gain being reduced to allow for bigger positive corrections. I think that's something to watch out for sure. In my case, my amps are also Krell and my speakers can handle the extra work in the LF without bottoming out or any extra audible distortion. With the new DSP gain function, you can also reduce the negative gain if the volume is too low. I prefer to keep it set at -8db. My DAC has plenty of headroom so it's no big deal in my case. But I think some folks have lower voltage DACs or preamps and need the extra gain.

I have also tried both the sofa and the seat 9 measurement configurations. I know some folks feel the sofa arrangement is better. I have to disagree. In my system, the 9 measurements in the cube formation in the seated position is best. The image is slightly more refined and accurate. The bass is slightly tighter. I have four filters loaded into the DAP so I can easily switch between different filters. I also created different filters for on axis and 90 degree measurements. My earthworks mic came with a fabulous calibration file which was only done on axis. Of course, I prefer the on axis filter over the 90 degree one used with the calibration file. I also tried it without the calibration file and didn't like the sound in the HF as much. This goes to show that even with one of the most accurate mics money can buy, you still need to use a good calibration file when doing these types of measurements.

Michael.

dallasjustice · Dec 20, 2013

Here are some bigger shots of the new DAP.

Whatmore · Dec 21, 2013

I'm very interested in this.
Can someone please let me know how difficult it is to use?

Amarra Symphony with Dirac Live Impulse Response Correction

WBF Founding Member

WBF Founding Member

WBF Founding Member

WBF Founding Member

WBF Founding Member

Banned

WBF Founding Member

WBF Founding Member

WBF Founding Member

Member Sponsor

WBF Founding Member

WBF Founding Member

Member Sponsor

Member Sponsor

WBF Founding Member

Member Sponsor

WBF Founding Member

Member Sponsor

Member Sponsor

Well-Known Member

Similar threads