Average of left/right not equal to left+right? - Home Theater Forum and Systems - HomeTheaterShack.com

Old 02-18-13, 10:56 PM Thread Starter
Shackster

Join Date: Jun 2012
Posts: 38
Average of left/right not equal to left+right?

I apologize if this has been asked before but I don't know how to search this - I took a room measurement with only the left channel speaker playing then another measurement with only the right channel speaker playing. I then created a plot averaging those two readings. Finally, I turned both speakers back on and let REW run its normal left-then-right routine.

The "average" plot looks completely different then the plot that REW produced running its normal left-then-right routine.

Please see attached file. "Average" is the manual average of the separate left/right measurements. "Both" is when I let REW run its normal routine (combines both channels).

Why are the two results different and which is actually correct when both speakers are playing?
Nonlinear is offline

Old 02-19-13, 12:37 AM
Shackster

Join Date: May 2011
Posts: 10
Re: Average of left/right not equal to left+right?

What is REW's normal left-then-right routine?
Sieglander is offline
Old 02-19-13, 02:36 AM
Elite Shackster

Join Date: Aug 2011
Location: New Zealand
Posts: 1,275
Re: Average of left/right not equal to left+right?

Hi i find the best way is to measure

Left main seperate
Right seperate
Then together (both)

By Averaging the two speakers what are you trying to achieve?

Normally i see that "Averaging" is more for measuring Multiple Listening Positions
Phillips is offline
Old 02-19-13, 12:21 PM Thread Starter
Shackster

Join Date: Jun 2012
Posts: 38
Re: Average of left/right not equal to left+right?

Quote:
Sieglander wrote: View Post
What is REW's normal left-then-right routine?
Oops, my mistake. I have been using DRC Designer also and it runs separate left/right sweeps.

REW 5 runs both speakers at the same time and generates 1 plot for the combination. Why, then, does it give you a choice of mono or stereo filter impulse export? Either way it's mono, right?
Nonlinear is offline
Old 02-19-13, 12:32 PM Thread Starter
Shackster

Join Date: Jun 2012
Posts: 38
Re: Average of left/right not equal to left+right?

Quote:
Phillips wrote: View Post
By Averaging the two speakers what are you trying to achieve?
I was wondering if it would give a better approximation of what I actually hear. The sharp nulls that show up in measurement plots may not be as audible as they look because we have two ears - only one ear may be in that null the other not.

Probably a better approach would be to run both speakers and then average two measurements taken at, say, 8 inches apart, to simulate our ear spacing. Still wouldn't be exactly right but might be a closer approximation to what we actually hear than a single point measurement or average of many random points.

This is for a mixing desk - relatively small working area in a medium sized room. I don't need the entire room flat just a relatively small "sweet spot" for proper mix balancing and EQ.
Nonlinear is offline
Old 02-19-13, 01:55 PM
Senior Shackster

Join Date: Feb 2013
Location: Sydney AU
Posts: 134
Re: Average of left/right not equal to left+right?

Hi, running both speakers together will result in destructive interference at different frequencies, whereas when doing an average you are looking at the magnitude only i.e. phase information is lost. The Nyal and Hedback white paper on acoustic measurement recommends running left and right separately above 250 Hz and both together below 250 Hz.

* http://blog.acousticfrontiers.com/wh...-rooms-pu.html
HifiZine is offline
Old 02-19-13, 02:37 PM Thread Starter
Shackster

Join Date: Jun 2012
Posts: 38
Re: Average of left/right not equal to left+right?

Quote:
HifiZine wrote: View Post
The Nyal and Hedback white paper on acoustic measurement recommends running left and right separately above 250 Hz and both together below 250 Hz.

* http://blog.acousticfrontiers.com/wh...-rooms-pu.html
Thank you for that info.

Now, how would one generate correction filters using this approach with REW? And how do you handle the transition region at 250Hz?
Nonlinear is offline
Old 02-20-13, 07:53 PM
Senior Shackster

Join Date: Feb 2013
Location: Sydney AU
Posts: 134
Re: Average of left/right not equal to left+right?

If you mean using the REW automatic filter generation, I'm not sure but perhaps the place to start would be to do the left and right independently, then measure both to see how the result looks < 250.
HifiZine is offline
Old 02-20-13, 10:22 PM
HTS Moderator
Reviewer

Wayne Myers

Join Date: Jun 2012
Location: Lincoln, NE
Posts: 4,609
Re: Average of left/right not equal to left+right?

Quote:
Nonlinear wrote: View Post
Thank you for that info.

Now, how would one generate correction filters using this approach with REW? And how do you handle the transition region at 250Hz?
It will be a partially manual operation (I have not tried this, but it should work):
• Generate filters for left speaker and save.
• Generate filters for right speaker and save.
• Run a measurement sweep with both speakers that only goes to 250 Hz.
• Generate low-frequency filters and save.
• Now manually inserted low-frequency filter values from the last step into the first two filter sets.
• Remeasure each speaker (& both together below 250) to verify the results.
• If the curves above and below 250 failed to line up, manually generate a house curve to compensate for the error and rerun the process.
AudiocRaver is offline
Old 02-21-13, 09:22 AM Thread Starter
Shackster

Join Date: Jun 2012
Posts: 38
Re: Average of left/right not equal to left+right?

Quote:
AudiocRaver wrote: View Post
It will be a partially manual operation (I have not tried this, but it should work):
• Generate filters for left speaker and save.
• Generate filters for right speaker and save.
• Run a measurement sweep with both speakers that only goes to 250 Hz.
• Generate low-frequency filters and save.
• Now manually inserted low-frequency filter values from the last step into the first two filter sets.
• Remeasure each speaker (& both together below 250) to verify the results.
• If the curves above and below 250 failed to line up, manually generate a house curve to compensate for the error and rerun the process.
OK, so basically what you are saying is to generate stereo correction filters above 250Hz and mono filters below 250Hz? Sounds like a lot of work that could create a real mess if not done properly! LOL

The only thing I'm not sure about in that case is the interaction above 250Hz when both speakers are playing the same tones. As I mentioned in one of my other posts here, I have been experimenting with "DRC Designer" which does create stereo correction filters using separate left and right sweeps. However, in my experience so far, the mono filters created by REW sound better. Probably because they account for the interactions.

Nevertheless, I will try out what you suggested and see what happens!
Nonlinear is offline

 Bookmarks

 Tags average , equal , left , left or right

Message:
Options

## Register Now

Random Question
Random Question #2

User Name:
OR

## Log-in

Human Verification

In order to verify that you are a human and not a spam bot, please enter the answer into the following box below based on the instructions contained in the graphic.