Very, very early impression. I listened to the 10 cycle and 2 cycle (I will get to the others in time). Wanted to get a sense for minimal vs maximum processing. About what I expected, the 2 cycle sounded better. By a good bit. The 10 cycle just seemed flat. Lost dynamics. Almost like a 192BR mp3.
More news later. May be a few days.
More news later. May be a few days.
I take it you have it on a CD? I'd say shuffle play it. It's not good to know what's playing.
You could easily fool yourself, really!
But I did not expect great things from the 10 cycle. But different things. It would also have a tiny sweetspot probably.
Shuffle play it, close your eyes and just listen over a few days. Pick it blind, don't peak. You'll know which version you like and then
look at the track number. That's the only way to get rid of the bias that everybody has up to some point.
You could easily fool yourself, really!
But I did not expect great things from the 10 cycle. But different things. It would also have a tiny sweetspot probably.
Shuffle play it, close your eyes and just listen over a few days. Pick it blind, don't peak. You'll know which version you like and then
look at the track number. That's the only way to get rid of the bias that everybody has up to some point.
Last edited:
I think you will prefer the track with either the 3 or 4 cycle window in the end. After a point, you are correcting things that (psychoacoustically speaking) don't need to be corrected. In other words, the music is combating room effects that either are not detracting from the sound or possibly even enhancing it. When you have the luxury of examining a number of filters, you will see there is a point of diminishing (and eventually negative) returns. With my system, I think this is around 4 cycles (5 still sounds very good, and with 6, the sweetspot is starting to shrink). If you want to better relate what you hear to what you see (at least for the range where our hearing is more sensitive), you could try windowing the graphs to ~8ms since this is probably the bulk of what we perceive as the direct sound.
I wouldn't be surprised if it were the 2 cycles version judging the FR plots.
The 10 cycles is down 7 dB in the bass compared to the 2 cycle version at 30 Hz. That alone can sound very different, but not be the fault of the number of cycles
.
Hard to make a truly controlled test as each version with more cycles of correction is down a bit in bass response compared to the one before it.
I'm more interested in which version makes instruments sound more real. But that could be a hard test with just one song. I hope the changes in target due to the number of cycles won't be the referee here.
I use different correction windows with less cycles in mid frequencies and don't want a long window on the bottom end as it does seem to rob some dynamics there. But on the top end I could go with quite a long window without objecting, as long as the mid window wasn't above ~4 cycles.
The differences were small at first, until I knew I hit a sweet spot and things improved dramatically. But I use way different speakers which may need more help in my room.
But it's not the room effects or lack there off. It was harmonics that really made it sing.
And atmosphere, like it is an acoustical event rather than a song you're listening to.
The 10 cycles is down 7 dB in the bass compared to the 2 cycle version at 30 Hz. That alone can sound very different, but not be the fault of the number of cycles
.Hard to make a truly controlled test as each version with more cycles of correction is down a bit in bass response compared to the one before it.
I'm more interested in which version makes instruments sound more real. But that could be a hard test with just one song. I hope the changes in target due to the number of cycles won't be the referee here.
I use different correction windows with less cycles in mid frequencies and don't want a long window on the bottom end as it does seem to rob some dynamics there. But on the top end I could go with quite a long window without objecting, as long as the mid window wasn't above ~4 cycles.
The differences were small at first, until I knew I hit a sweet spot and things improved dramatically. But I use way different speakers which may need more help in my room.
But it's not the room effects or lack there off. It was harmonics that really made it sing.
And atmosphere, like it is an acoustical event rather than a song you're listening to.
Yes, I noticed the difference is bass output as well. I think that is because the bass with the 2 cycles correction is still "booming" a bit (something I happen to have a small tolerance for). I would take rolled-off bass any day over booming. In fact, my tiny speakers have forced me to learn what is the minimum bass response that still delivers the sound convincingly. For me, that would be a first order slope that is -3db at 64hz (flat to 128hz and -9db at 32hz) and within 90 deg min phase at 32hz. I digress...
You make a good point about judging the sound based on the target FR more so than the time domain behavior. It's tricky though because they're intertwined.....
You make a good point about judging the sound based on the target FR more so than the time domain behavior. It's tricky though because they're intertwined.....
One persons "boom" is anothers "boon"
I dont think the FR is the only thing involved in this perception. I find I have a higher tolerance for elevated bass when its tight than when its not. I think it also depends on what genre of music im listening to.
I like rock with a firm kick and strong bass guitar. I like to feel the impact in the bass region. Not to the point where it masks the lower mids, but maybe just short of that.
I did notice the elevated bass in the 2 cycle FR curves. But its mainly limited to below 40hz, where most music I hear has little output. So I didnt notice it when listening earlier.
To be picky, that's not quite right, I think you meant "gives zero GD variation".
To further embellish the discussion and further confuse anyone puzzled by phase and delay: A constant time delay isn't the same as constant phase, it just means the the phase is everywhere proportional to frequency (phase is =SomeMagicConstant*frequency). On a linear frequency scale, the phase response will have a constant slope. On a log-frequency scale the phase curve slope will increase continuously (get more steep downwards) with frequency. With linear phase response data, you can remove some fixed amount of delay to arrive at a phase response that is a flat line at 0 degrees on a linear or log-frequency graph.
Fun fact: flat group delay, or even 0 seconds group delay, doesn't guarantee waveform fidelity, even if the magnitude (dB) response is also flat. For instance, if a phase response were to be 90 degrees at all frequencies, any waveform going through that would be extremely distorted but the group delay would be 0 at all frequencies! For faithful waveform replication, the constant group delay phase response has to be one in which, when the magic value of delay is remove to make the phase line flat, that phase value is 0degrees. (If 180 degrees, just reverse some wires)
Last edited:
One persons "boom" is anothers "boon"
I dont think the FR is the only thing involved in this perception. I find I have a higher tolerance for elevated bass when its tight than when its not. I think it also depends on what genre of music im listening to.
I like rock with a firm kick and strong bass guitar. I like to feel the impact in the bass region. Not to the point where it masks the lower mids, but maybe just short of that.
I did notice the elevated bass in the 2 cycle FR curves. But its mainly limited to below 40hz, where most music I hear has little output. So I didnt notice it when listening earlier.
True. In the frequency range that really matters for music, there is really no boom at all. But if you want time domain accuracy down to the sub-bass range, you can achieve it using 3-4 cycles of correction. That is quite an accomplishment.
An accomplishment technically. But I am still trying to decide if it matters sonically.
To be picky, that's not quite right, I think you meant "gives zero GD variation".
To further embellish the discussion and further confuse anyone puzzled by phase and delay: A constant time delay isn't the same as constant phase, it just means the the phase is everywhere proportional to frequency (phase is =SomeMagicConstant*frequency). On a linear frequency scale, the phase response will have a constant slope. On a log-frequency scale the phase curve slope will increase continuously (get more steep downwards) with frequency. With linear phase response data, you can remove some fixed amount of delay to arrive at a phase response that is a flat line at 0 degrees on a linear or log-frequency graph.
Fun fact: flat group delay, or even 0 seconds group delay, doesn't guarantee waveform fidelity, even if the magnitude (dB) response is also flat. For instance, if a phase response were to be 90 degrees at all frequencies, any waveform going through that would be extremely distorted but the group delay would be 0 at all frequencies! For faithful waveform replication, the constant group delay phase response has to be one in which, when the magic value of delay is remove to make the phase line flat, that phase value is 0degrees. (If 180 degrees, just reverse some wires)
Very informative post! Thanks!
I don't think this applies to multiways with a XO, as I have personally seen flat FR but phase is not.
You were right. The kicker was causing the extra GD at 11k. I am glad this got addressed because I discovered with the new high end FR, the kicker was only -2db at 11k! A bit too much kick there I think, so I brought it back down to about -5db.
Glad you found that source Jim, hope you're enjoying the test with the different correction strengths. I kind of get the feeling you're looking for a way to get the same results acoustically though (lol). Nothing wrong with that, the more you fix acoustically the better it gets. And you've done a very impressive job there already. DRC almost seems like cheating in comparison.
I removed the infrasonic BW2 filter on my FAST and got the entire GD to be +/-2.5ms above 33Hz. Does this make sense? It seems to still sound good but I do see the cone moving more.
http://www.diyaudio.com/forums/full...-rs225-8-fast-ref-monitor-43.html#post4342580
http://www.diyaudio.com/forums/full...-rs225-8-fast-ref-monitor-43.html#post4342580
Your partly right. What i am doing is isolating the things that DRC does that I like (Bass GD, FR tilt and channel balance) and adjusting my room acoustically.
I dont expect, nor am I attempting to match, all that DRC does.
Some of the improvements DRC offers I think I can accomplish acoustically, yes.
A bit off topic, but one of the things that DRC showed me was better image location clarity. It got me to thinking about what factors contribute most to great imaging in a fundamental way.
One of the things DRC did well was even the channel balance, and perhaps some of the phase miss matches.
Even timing and magnitude, one channel compared to the other, IMO, are two of the most important areas contributing to image clarity.
In looking at my previous phase and FR plots, I found some unevenness that I could fix acoustically, and with speaker XO tweeks.
Especially in the tweeter range, but some midrange as well, closer phase and magnitude responses did improve imaging audibly.
At 1/24th oct smoothing, most of my deviation is within 1db (250hz - 4K). A couple isolated areas up around 1.6db. The DRC 4 cycle version brings it all down to under 0.9db.
Channel Deviation (observed results)
--------------------- Average/Peak
DRC ------------------- 0.5/0.9db
Now (no DRC) --------- 1/1.6db
Before (no DRC) -------1.5/2.5db
One of the things DRC did well was even the channel balance, and perhaps some of the phase miss matches.
Even timing and magnitude, one channel compared to the other, IMO, are two of the most important areas contributing to image clarity.
In looking at my previous phase and FR plots, I found some unevenness that I could fix acoustically, and with speaker XO tweeks.
Especially in the tweeter range, but some midrange as well, closer phase and magnitude responses did improve imaging audibly.
At 1/24th oct smoothing, most of my deviation is within 1db (250hz - 4K). A couple isolated areas up around 1.6db. The DRC 4 cycle version brings it all down to under 0.9db.
Channel Deviation (observed results)
--------------------- Average/Peak
DRC ------------------- 0.5/0.9db
Now (no DRC) --------- 1/1.6db
Before (no DRC) -------1.5/2.5db
Attachments
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.