ProSoundWeb Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: 1 [2] 3 4   Go Down

Author Topic: question for the pro's on Comb Filtering  (Read 14447 times)

Ivan Beaver

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 9031
  • Atlanta GA
Re: question for the pro's on Comb Filtering
« Reply #10 on: September 09, 2012, 08:04:06 pm »

Now here is an embarassing situation in which basic combfiltering bit me.

A couple of years ago we had a customer who was looking at buying either a Yamaha PM1D or a Innovason SYS80.

So we had both of them setup side by side at the shop, for them to do a sound quality and feature/operation side by side test.

I had used the Yamaha earlier in the week to do a 5 day festival.  Remember that -it will be important later.

So we used a Neuman mic and a Frazier speaker for some basic mic tests.

So I bypassed all the eq-processing etc on both consoles and routed channel 1 directly to the main output-which went ot the amp and speaker.

We tested the Yamaha first-and it sounded a little bit "dull"  Not bad-but I was unfamiliar with sound of both the mic and the speaker and didn't think a lot of it.

So we switched to the Innovason.  WOW!!!!!! It was soooooo much clearer-open-detailed etc.  We went back to the Yamaha and I doubled checked everything on the channel-master-noting was inserted and so forth.

But the Innovason just sounded soooo much better.  I felt that something was wrong-but could not put my finger on it.

The Innovason rep said the difference was in the mic pres-and the Innovason simply had better preamps.

This demo was at night.  We were having an in house Syn Aud Con at the time.  The next day I opted out of the class to try and figure out what was going on.

I could not find anything that could be wrong.  So when all else fails MEASURE!  I hooked up Smaart to the PM1D and did an impulse response and HOLD THE HORSES! I was seeing TWO arrivals!  A transfer function measurement showed the classic comb filter notches.

BUT WHERE WAS THE PROBLEM?????????

Remember that Festival I did?  Well we were using channel 1 for the test mic.  That was the problem.  For the festival I used channel 1 for the kick mic.  I had that mic routed to a drum subgroup for the show.

So when I assigned it directly to the main output-I had TWO parths for the signal going to the mains.  No big deal in analog world.  But on this (and many other digital desks) this is a NO NO.  I was unaware of that at the time.  There was additional latency (less than 1ms).  yes the "little" differences can be a real pain.

As soon as I stopped one of the paths (it didn't matter which one)-the sound cleared right up.  SOOOO much better.

NOW the console sounded like I thought it should.

I interuppted the class and wanted to use this as a classis demonstration of comb filtering and how its presence (and this was worse case and technically "perfect") can be hard to track down.  It fooled me for awhile.

When we went between the Yamaha and the Innovason-the sonic differences were now non existant.  At least in the casual testing we were doing.

However the Innovason rep had been there the night before-and was unaware of what I had found.  I am quite sure he has told many people how he did a side by side with a PM1D and the Innovason clearly won in sound quality.

I would agree-in the particular situation-but in a way-we were dealing with a crippled PM1D-due to my lack of knowledge about double bussing on digital consoles.

SOrry for the long story-but I feel it needs to be told-if for no other reason a real life situation that does crop up and to be used for education for others.
Logged
A complex question is easily answered by a simple-easy to understand WRONG answer!

Ivan Beaver
Danley Sound Labs

PHYSICS- NOT FADS!

Merlijn van Veen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 309
    • www.merlijnvanveen.nl
question for the pro's on Comb Filtering
« Reply #11 on: September 09, 2012, 08:10:21 pm »

A 2 speaker coupled point source, say 2 times 60 degrees at 60 degrees splay, might exhibit pattern narrowing in the lower mids at half wavelenght displacement. A 3 speaker coupled point source, say 3 times 40 degrees at 40 degrees splay, yields the same coverage angle but most likely with more a uniform pattern in return for 2 acoustic crossovers instead of 1. Pick your poison.

Doug Fowler

  • Member since May 1995, 2nd poster on original LAB, moderator on and off since 1997, now running TurboMOD v1.826
  • Administrator
  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1769
  • Saint Louis, MO USA
Re: question for the pro's on Comb Filtering
« Reply #12 on: September 09, 2012, 08:22:18 pm »

Now here is an embarassing situation in which basic combfiltering bit me.

A couple of years ago we had a customer who was looking at buying either a Yamaha PM1D or a Innovason SYS80.

So we had both of them setup side by side at the shop, for them to do a sound quality and feature/operation side by side test.

I had used the Yamaha earlier in the week to do a 5 day festival.  Remember that -it will be important later.

So we used a Neuman mic and a Frazier speaker for some basic mic tests.

So I bypassed all the eq-processing etc on both consoles and routed channel 1 directly to the main output-which went ot the amp and speaker.

We tested the Yamaha first-and it sounded a little bit "dull"  Not bad-but I was unfamiliar with sound of both the mic and the speaker and didn't think a lot of it.

So we switched to the Innovason.  WOW!!!!!! It was soooooo much clearer-open-detailed etc.  We went back to the Yamaha and I doubled checked everything on the channel-master-noting was inserted and so forth.

But the Innovason just sounded soooo much better.  I felt that something was wrong-but could not put my finger on it.

The Innovason rep said the difference was in the mic pres-and the Innovason simply had better preamps.

This demo was at night.  We were having an in house Syn Aud Con at the time.  The next day I opted out of the class to try and figure out what was going on.

I could not find anything that could be wrong.  So when all else fails MEASURE!  I hooked up Smaart to the PM1D and did an impulse response and HOLD THE HORSES! I was seeing TWO arrivals!  A transfer function measurement showed the classic comb filter notches.

BUT WHERE WAS THE PROBLEM?????????

Remember that Festival I did?  Well we were using channel 1 for the test mic.  That was the problem.  For the festival I used channel 1 for the kick mic.  I had that mic routed to a drum subgroup for the show.

So when I assigned it directly to the main output-I had TWO parths for the signal going to the mains.  No big deal in analog world.  But on this (and many other digital desks) this is a NO NO.  I was unaware of that at the time.  There was additional latency (less than 1ms).  yes the "little" differences can be a real pain.

As soon as I stopped one of the paths (it didn't matter which one)-the sound cleared right up.  SOOOO much better.

NOW the console sounded like I thought it should.

I interuppted the class and wanted to use this as a classis demonstration of comb filtering and how its presence (and this was worse case and technically "perfect") can be hard to track down.  It fooled me for awhile.

When we went between the Yamaha and the Innovason-the sonic differences were now non existant.  At least in the casual testing we were doing.

However the Innovason rep had been there the night before-and was unaware of what I had found.  I am quite sure he has told many people how he did a side by side with a PM1D and the Innovason clearly won in sound quality.

I would agree-in the particular situation-but in a way-we were dealing with a crippled PM1D-due to my lack of knowledge about double bussing on digital consoles.

SOrry for the long story-but I feel it needs to be told-if for no other reason a real life situation that does crop up and to be used for education for others.



Some years back when Scovi was still using analog, he said he would not switch until there was bus compensation re: latency on a digital desk. Of course he has been with Digi (or whatever :-) since and it's not an issue.

In my SysTune classes we spend time listening to comb filter examples with both speech and music.   The fact is many don't know what to listen for, until they are shown.

It's an ear-opening experience for many.

I have found that by using noise for system alignment tasks, it is much easier to hear delay offsets than with music.
Logged
"It's got electrolytes.  It's got what plants crave."

Sharyn Ferrick

  • Newbie
  • *
  • Offline Offline
  • Posts: 31
Re: question for the pro's on Comb Filtering
« Reply #13 on: September 09, 2012, 10:00:33 pm »

A 2 speaker coupled point source, say 2 times 60 degrees at 60 degrees splay, might exhibit pattern narrowing in the lower mids at half wavelenght displacement. A 3 speaker coupled point source, say 3 times 40 degrees at 40 degrees splay, yields the same coverage angle but most likely with more a uniform pattern in return for 2 acoustic crossovers instead of 1. Pick your poison.

Thanks, I am sort of limited in that the EV zxa5's only have two models 60x60 and 90x60 so it looks like two 60x60 makes the most sense.  I guess another options would be to have 8 of them mounted 2x2 splayed 60+60 so it would be 120 vertical and 120 horizontal, but my thinking is I am better with using the dual delays, and leaving the fronts to just dual 60x60 on each side
Sharyn
Logged

Sharyn Ferrick

  • Newbie
  • *
  • Offline Offline
  • Posts: 31
Re: question for the pro's on Comb Filtering
« Reply #14 on: September 09, 2012, 10:03:32 pm »



Some years back when Scovi was still using analog, he said he would not switch until there was bus compensation re: latency on a digital desk. Of course he has been with Digi (or whatever :-) since and it's not an issue.

In my SysTune classes we spend time listening to comb filter examples with both speech and music.   The fact is many don't know what to listen for, until they are shown.

It's an ear-opening experience for many.

I have found that by using noise for system alignment tasks, it is much easier to hear delay offsets than with music.
For those of us in remote areas and not able to attend any of your classes, would a series of .wav files be able to teach what to listen for?

Thanks
Sharyn
Logged

Ivan Beaver

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 9031
  • Atlanta GA
Re: question for the pro's on Comb Filtering
« Reply #15 on: September 10, 2012, 07:33:51 am »

For those of us in remote areas and not able to attend any of your classes, would a series of .wav files be able to teach what to listen for?

Thanks
Sharyn
If you have access to a digital console that has delay on the inputs-it is real easy to do.

Just take a single input and assign it to 2 channels.  Set the gain the same.

Turn up one channel and listen to the main output.  Now turn up the other.  THe level should increase 6dB (if all things are set the same).

Now add a little bit of delay (say 1ms) to one of the inputs.  Hear the sound change?  THAT is combfiltering.  Now play with the delay time and see how it changes.

If you only have delay on the outputs-then take 2 outputs and sum them together in a DSP or other mixer.  You could even return the 2 outputs to 2 inputs on the console-assign those 2 inputs to a DIFFERENT output (say an aux) and listen to the aux.
Logged
A complex question is easily answered by a simple-easy to understand WRONG answer!

Ivan Beaver
Danley Sound Labs

PHYSICS- NOT FADS!

Joseph D. Macry

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 421
  • Austin TX
Re: question for the pro's on Comb Filtering
« Reply #16 on: September 10, 2012, 11:47:56 am »

Remembering a scene from TV series "24": Jack Bauer has tracked down some audio guy who has altered a forensic audio file to hide evidence. It came down to how he managed to remove a digital watermark. "I just apply the comb filter here, like this..."   ::)
Logged
Joseph Macry,
Austin, TX

John Roberts {JR}

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 16552
  • Hickory, Mississippi, USA
    • Resotune
Re: question for the pro's on Comb Filtering
« Reply #17 on: September 10, 2012, 11:58:34 am »

Remembering a scene from TV series "24": Jack Bauer has tracked down some audio guy who has altered a forensic audio file to hide evidence. It came down to how he managed to remove a digital watermark. "I just apply the comb filter here, like this..."   ::)
To continue this veer, and coincidentally, at one point there was a proposal from CBS for a digital audio watermark that was an extremely narrow notch filter. The CBS argument was that it would be so narrow as to be inaudible to consumers, but reliable for digital ID purposes.

Thankfully it failed to gain traction.

JR
Logged
Don't tune your drums half-ass. Listen to what a properly "cleared" drum sounds like.   http://circularscience.com/

Mike Caldwell

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1788
  • Covington, Ohio
    • Mike Caldwell Audio Productions
Re: question for the pro's on Comb Filtering
« Reply #18 on: September 10, 2012, 05:43:02 pm »

To continue this veer, and coincidentally, at one point there was a proposal from CBS for a digital audio watermark that was an extremely narrow notch filter. The CBS argument was that it would be so narrow as to be inaudible to consumers, but reliable for digital ID purposes.

Thankfully it failed to gain traction.

JR

I remember reading about that "brilliant" idea. Actually reading the arguments for it were amusing, most of it had to do with trying to prevent digital to digital copying.

Merlijn van Veen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 309
    • www.merlijnvanveen.nl
question for the pro's on Comb Filtering
« Reply #19 on: September 10, 2012, 06:33:06 pm »

Taking it one step further, after listening to comb filtering with pink noise or music. Use a repetitive broad spectrum noise impulse signal, like a metronome. Not only do you hear the onset of comb filtering but also, when enough delay is added, echo's. Which turn out to be frequency dependent. Quite contrary to most literature that says that echo's start around 60 ms, regardless of frequency. I found this quite revealing. Apparently it's more likely to state that echo's occur at 24 wavelengths of delay or more. AFAIK credit for this little experiment, in my case, goes to Bob McCarthy.

ProSoundWeb Community

question for the pro's on Comb Filtering
« Reply #19 on: September 10, 2012, 06:33:06 pm »


Pages: 1 [2] 3 4   Go Up
 



Page created in 0.036 seconds with 22 queries.