Topic: Brickwalled recordings

[Ekib]

Some of my older recordings , like 1991-1992 etc. are brickwalled. I didn't use the -20 DB settings at every concert.
They're not really bad , most of the time they're listenable. It's just when there is a really high note ( like a guitarsolo ) you can hear it.
Now , is there a trick to get rid off this ? Is there some kind of software that can fix this ?

Thanks !

[boojum]

Samplitude offers a clipping plugin that can help some.  I am sure other editors can, too.  Audacity has more plugins than Carter has pills, so there must be one for this.  I'd check it out if I were you.  Audacity and the plugins are free.

[DSatz]

Adobe Audition also has a "de-clipping" feature that work fairly well with mild to moderate clipping (say, up to 3 - 4 dB). If you try to fix a recording that has 10 dB of clipping or more (i.e. the peak levels would have been 10 dB greater if they hadn't been clipped), the software's projections become less appropriate and the result sounds noisy and strange.

An old friend of mine once called this the "law of conservation of goodness." From experience it seems to be a general rule--if you can fix a minor problem with a small amount of processing, it's far more likely to succeed than trying to fix a major problem with a lot of processing.

[newplanet7]

What about brickwalling that occurs before the clipping threshold?
I was under the impression that clipping and brickwalling were two different entities.
Clipping is actually the preamps being overloaded and brickwalling was
the preamp overloading the ad stage thus no actual clipping (0db) has to be present.

My first time out I bricked and my levels were around -4 tops the whole time.
Am I misunderstanding this? 

[newplanet7]

^^^^^^^^^^^^^^^^^^^^^
Bueller.............. Bueller.............. Bueller...............

[DSatz]

Any time a signal runs into a hard limit, that's clipping. Regardless of what causes it, the term still applies.

"Brickwalling" is a term that I frankly never heard before I came to this forum. It's not in general use in professional audio; "clipping" is the term that everyone (outside of here) knows.

[Gutbucket]

Beat me to it..

Brickwalling, clipping.. same difference. You run out of headroom and the waveform is flattened at the peaks and troughs.  Around these parts people call it brickwalling when a previous stage clips hard somewhere before the meters.  In that case the levels on the meters look OK but the signal is already clipped before it gets to the ADC.  Some analog stages may distort more gently when they are slightly overloaded, squashing the waveform in a more gradual, rounded way. A small amount of that distortion can sometimes sound good, but they can also clip sharply as if hitting a proverbial brick wall, sounding ugly.  I've never heard the term brickwalling term outside of this forum.

[newplanet7]

Exactly what I was getting at.
I can "clip" at -4db where the waveform is boxed off at -4db.
Most people around here refer to clipping as "going over" 0 right?

Not too many people see clipping as anything but going over 0.
Thus the term brickwalling was born?
I am still confused.
I'd like to hear more people weigh in on this because
a lot of threads here treat it as two separate things.

My basic point being that clipping is in one part of the signal chain and brickwalling
is at a latter part of that signal chain
Thus both being different.
I guess My main question is how do you clip without coming close to "over"

[fmaderjr]

DSatz could explain the technicalities if he has time.

In practice, most (all?) of the recorders we use have a setting on the level control that if you need to turn the level below that point to keep the meters from going over 0 dB, you are likely to have brickwall distortion. If you learn this point for your recorder, you should never have a problem with brickwall distortion (as long as the sound pressure isn't more than your mics can handle). Instead of turning your levels below this point you can use a pad or a preamp capable of attenuation (like the ST-9100).

guysonic mentions some of these in the recorder reviews on his site.

Some of these points I've read about on this forum:
R-09             8
MZ-RH1        12
iHP-120       -0.5
Korg MR-1   -10
 

[Roving Sign]

Brickwalling is certain kind of clipping - referring specifically to the overload of the first input stage on any given recorder. No amount of gain adjustment will clean up the signal.

Example: The old Sony TCD5M Analog cassette field recorders had a -20db pad switch. You were guaranteed to distort(brickwall) the signal if you did not use this pad.

Clipping is more like when you set your levels too high and overdrive something downstream. (tape head/A>D converter)

[Roving Sign]

Beat me to it..

Brickwalling, clipping.. same difference. You run out of headroom and the waveform is flattened at the peaks and troughs.  Around these parts people call it brickwalling when a previous stage clips hard somewhere before the meters.  In that case the levels on the meters look OK but the signal is already clipped.

And to that point - Once you screw up a few times you can get an eye for meters that dont "look right"

You'll notice a distinct lack of bounce to your signal meters - only the top few bars will blink.

VU needles are more revealing...no swing. Just a trembling needle.

[newplanet7]

Thanks fellas.
It only happened to me one time/first time

[Gutbucket]

My basic point being that clipping is in one part of the signal chain and brickwalling
is at a latter part of that signal chain
Thus both being different.
I guess My main question is how do you clip without coming close to "over"

To rehash, Brickwalling is a provincial term only really used around here, but as used here you have it backwards.  It is clipping earlier in the signal chain than the point where the meters are monitoring. 

Because levels can be adjusted between stages it is possible to clip the input stage then have the signal level reduced to a manageable level before reaching the later stage where you are monitoring the meters.  Like Roving mentions, when this occurs you may notice that the level meter isn't 'bouncing' right, but just tops out at a certain point below full scale.  That's a subtle indication that the clipping is occurring earlier in the chain.  If there was a separate calibrated meter at that earlier stage, you would see it overloading, but by the time the signal gets to the final stage with the meters, the signal level is back under control.  The recorder is at that point faithfully recording an already clipped signal, regardless of where you set the final levels.

Instead of an all-in-one recorder, imagine a signal chain with separate components for each function: mics > mic preamp > ADC > digital recorder.  Imagine separate level adjustments at the mics (attenuating pad switch), preamp (gain) and ADC (gain/attenuation).  If there were also meters on each component, you could adjust everything so that each component was operating in a comfortable range.  Going though that process is called gain-staging and is an important part of setting up a studio recording chain.  In an all-in-one recorder all those separate boxes are in one device, and we rely on the manufacturer to optimize the gain-staging for us since they are condensing all that stuff into one device. 

[newplanet7]

  Because levels can be adjusted between stages it is possible to clip the input stage then have the signal level reduced to a manageable level before reaching the later stage where you are monitoring the meters.  Like Roving mentions, when this occurs you may notice that the level meter isn't 'bouncing' right, but just tops out at a certain point below full scale.  That's a subtle indication that the clipping is occurring earlier in the chain.  If there was a separate calibrated meter at that earlier stage, you would see it overloading, but by the time the signal gets to the final stage with the meters, the signal level is back under control.  The recorder is at that point faithfully recording an already clipped signal, regardless of where you set the final levels.

And there it is. 
That makes perfect sense.
Thanks gutbucket/roving.

[it-goes-to-eleven]

To rehash, Brickwalling is a provincial term only really used around here, but as used here you have it backwards.  It is clipping earlier in the signal chain than the point where the meters are monitoring. 

An example where it doesn't happen earlier in the chain is a case like the r09 being run at trim settings below 8.  It can brickwall without clipping, and before -6.

[Gutbucket]

Well that's really no different.  It's still happening earlier in the chain, its just that most or all of the chain is in one device.  In the case of any of these all-in-one recorders you have what used to be multiple boxes (preamp, ADC, recorder.. mics too) all in one unit.  When brickwalling happens in the R-09, the analog input preamp stage of the chain within recorder is clipping.

Of course in that case, we have no way of changing the gain staging between the components within the device itself.  That's Edirol's job in setting up the gain-staging when they designed the machine.  It can be strongly argued that allowing the user to set input levels below level 8 on that device is a design flaw.  After all they have all the components under their control.  But once we know not to use such low settings it's not a big deal. We just stay away from that region.  Similarly I don't run the gain up too high because noise becomes a problem.  I mentioned this elsewhere, but my comfort region on the R-09 is about 10-20 or so.  If I need more gain then that I use an external preamp or a sensitive mics.  If I needed less I'd use attenuators before the recorder's inputs or less sensitive mics.

[newplanet7]

So basically this is less likely to happen with separate components
rather than all in ones?
Basically because you get to monitor each stage?

[Gutbucket]

If you had actually had meters everywhere that would be true, but there usually isn't metering available on every unit.  Take pre-amps for example.  I have meters on the V3 but none on the MMA6000 or my Church Audio pres.  If I run the V3 in front of the R-09 I can check levels on both of those units.  Because the R-09 doesn't accept a very hot signal, the V3's meters only light up the first few lights when I set the r-09 input at 10 (the lower end of my comfort range).  That's OK, nothing is overloading, but I don't really need the meters on the V3 in that case.  It's sort of the opposite of the brickwalling, the V3 has way more headroom than needed in this case and is in no danger whatsoever of clipping.  But the V3 is a more professional piece of gear designed for higher levels and with more headroom in its analog sections.

I've thought about doing the math to figure out at what point I will be close to clipping those other preamps so that I actually know what's going on with the gain staging in my chain.  I know the sensitivity of the mics and have specs on some of the preamps, but the math part is a bit beyond my skills and I can comfortably run my gear without knowing exactly so doing that would be a bit academic.  That's something I'd need DSatz or one of the EE's on this board to help me through.

In actuality, all-in-one recorders should be less likely to have problems because the manufacturer is responsible for making sure all the components work together at appropriate levels.  Yet cost constraints manufacturing inexpensive recorders means that there is no budget for extra headroom for unusual 'just in case' scenarios and these types of problems can manifest with extreme settings.

[Ekib]

That's nice and all ...the difference between brickwall and clipping...thanks for the info...but...my question was is there any way to correct this with any type of software ? I have anto clipping in for example Nero Wave Editor and all it does is lower the volume. Not the distortion itself.
I am just wondering if there is any way to fix it.
A friend of mine compares it to scratches on a picture ...whatever isn't there can't be corrected he says . But I think it is the opposite. To me it looks like there is too much info ( for lack of a better word ). It's not like something's missing , it's just too much.

[newplanet7]

I don't think you'll be able to undo the distortion/saturation.
When I bricked I just deleted it.

[it-goes-to-eleven]

You can use the pencil tool to fix it by hand.   You might be able to use an expander to spruce things up but that's iffy.  Maybe in 5 or 10 years they'll be able to cure it with stem cells.  The fix everything!

Sometimes you have to accept it as a learning lesson and move on.   As long as the artist hasn't died and this was their last performance, etc, you'll get another chance.

[Gutbucket]

That's nice and all ...but...my question was is there any way to correct this with any type of software ?

A qualified yes.. see the first two responses to this thread.

[jlykos]

Sony Sound Forge has something called the Clipped Peak Restoration, which appears to me to be a kind of compression where it take s the signal and lowers it to a certain extent and then cuts it off at a user-defined limit.  I used it on a recent Mastodon recording where I accidentally knocked the -20 db switch to 0 and it improved the recording to a significant extent, but it does sound quite different (i.e. compressed) after the processing.

[Gutbucket]

I've only noticed the Samplitude version since that's the software I use, but haven't used the function.  All of the tools mentioned probably lower the level of the signal a bit to 'make enough room' to recreate the peaks, then use some sort of algorithm to guess the shape of the wave that should be there if it wasn't flat topped by clipping..  Sort of like enlisting a bevy of ompaloompas with pencil tools to redraw the tops and bottoms of the peaks like Freelunch mentioned.

Like most things the less you need it the better the results are likely to be, ala the 'conservation of goodness' law.

[newplanet7]

But would this really minimize the distortion?
I don't see how it could.
It seems that it would just lower the volume of the distortion.
I know when I bricked Sound Forge didn't do squat.

[jlykos]

But would this really minimize the distortion?
I don't see how it could.
It seems that it would just lower the volume of the distortion.
I know when I bricked Sound Forge didn't do squat.

My relative success with the Clipped Peak Restoration may have also been a function of using the limiter on the D50, which kicks in when the signal hits 0 and then reduces it around -12 db.  As a result, the distortion caused by clipping is not nearly as high as on other recorders.  There is still distortion present, particularly on the bass, but the recording sounds OK otherwise, even though it is quite compressed.

[newplanet7]

thanks jaime.
Now when you say clipped, you mean 0 right?
That, to my knowledge is the only way it will work in SF
because it looks for anything @ 0db.
I may be wrong though.

When I "bricked" I didn't have any peaks above -4.
Just squared off waveforms at -4db that were distorted.
Since then though I just run the pad on my 660 and all is well.
-todd

[jefflester]

They term "brickwalling" was used for years on DAT-heads, so not just around here. Though I suppose one could say the "taper community" which includes DAT-heads and TS. It was a problem with the Sony TCD-D7 and D8. The Sony TCD-D3 was the earlier generation small Walkman sized model and it was more tolerant of input level, the D7 came along as the next generation but was less tolerant.

Here's a vintage DAT-heads post describing the issue:
http://www.rockpark.com/d7/8_brick.txt
 

[newplanet7]

They term "brickwalling" was used for years on DAT-heads, so not just around here. Though I suppose one could say the "taper community" which includes DAT-heads and TS. It was a problem with the Sony TCD-D7 and D8. The Sony TCD-D3 was the earlier generation small Walkman sized model and it was more tolerant of input level, the D7 came along as the next generation but was less tolerant.

Here's a vintage DAT-heads post describing the issue:
http://www.rockpark.com/d7/8_brick.txt
 

thanks for the article link.

[Ekib]

They term "brickwalling" was used for years on DAT-heads, so not just around here. Though I suppose one could say the "taper community" which includes DAT-heads and TS. It was a problem with the Sony TCD-D7 and D8. The Sony TCD-D3 was the earlier generation small Walkman sized model and it was more tolerant of input level, the D7 came along as the next generation but was less tolerant.

Here's a vintage DAT-heads post describing the issue:
http://www.rockpark.com/d7/8_brick.txt
 

I have brickwalled recordings made with the TCD-D3 aswell.
Can't even remember how much battery trouble I've had with this recorder ! It broke down so many times.
I was glad I ws able to switch to the TCD-D7 and be able to use penlites...

[zzrck]

i thought brickwalling was where the mic was overloaded on the sound level.
i taped a ted nugent show , the recording level was set low , it didnt even come close , but the sound level itself
was way higher then my mic could handle , so even tho the recording level on my recorder was
set on the low side , the recording itself was bigtime distorted couse the mic was overloaded .
there was nothing i could do to save the recording ,  i finally gave up on trying to save it.

[jlykos]

thanks jaime.
Now when you say clipped, you mean 0 right?
That, to my knowledge is the only way it will work in SF
because it looks for anything @ 0db.
I may be wrong though.

When I "bricked" I didn't have any peaks above -4.
Just squared off waveforms at -4db that were distorted.
Since then though I just run the pad on my 660 and all is well.
-todd

Sorry, I misread your original post.  If you brickwalled the preamp, there is very little that you can do about the recording in my experience.  Clipped is going over 0 on the A/D.  The pad on the 660 will help with clipping if you have mics > 660, but if you run a preamp in front of it, there is nothing that the 660 can do about what happens at the preamp stage.

Good luck with everything in the future!

[newplanet7]

No problem.
It only happened once in 2008 when I first got it.
Just because I didn't run the pad on  the recorder.
Everything sense then has been great.