What does th linear phase eq do that for example a graphic eq doesn't
do? Ultimately, what audible difference will it make on a track or
mix? If I created an envelope with the graphic that looked exactly
like the linear phase eq, why would they sound different?
Thanks.
> Apparently a linear phase eq is considered an essential tool,
by some salesman perhaps.
> apparently Soundforge 5 doesn't have one - it has graphic, paragraphic
> and parametric. And to be honest I'm not altogether clear as to why
> these are different from each other either. The paragraphic seems like
> 4 parametrics bundled together, but I assume it's more than that.
> What does th linear phase eq do that for example a graphic eq doesn't
> do?
It alters the frequency response curve without altering the phase response
curve. That is a fine concept except that most aberrations you will end up
eq-ing are likely to be at least to some extent minimum phase, which is to
say that total phase response is more likely to be improved by an ordinary
"minimum phase" eq than by a "linear phase" eq.
> Ultimately, what audible difference will it make on a track or
> mix? If I created an envelope with the graphic that looked exactly
> like the linear phase eq, why would they sound different?
Because one alters the relative arrival time for components of differing
frequencies and the other doesn't. Expect this to be about subtleties. There
may be some advantage if uu aim for a grossly un-natural sound, but in that
context it probably doesn't matter anyway.
That's my opinion on this.
> Thanks
Kind regards
Peter Larsen
In the real world, all equalizers are linear phase unless something very
dramatic has been done to change that. What folks really mean is that
the system is "minimum phase" and correspondingly the phase shift is
coupled to the amplitude change in a standard and predictable way.
>what does th linear phase eq do that for example a graphic eq doesn't
>do? Ultimately, what audible difference will it make on a track or
>mix? If I created an envelope with the graphic that looked exactly
>like the linear phase eq, why would they sound different?
Your graphic EQ is probably already minimum phase. However, you will
find that a graphic EQ is an almost unbearably blunt tool that isn't
much good for anything, and that a parametric is a lot more powerful
as well as a lot more controllable.
--SCott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
It screws up things even worse.
Most frequency response errors include phase errors that are directly
related to the amplitude errors. (This is the way "nature" normally works.
See Hilbert.) An equalizer should include complementary phase errors, to
cancel them out. Because linear phase removes the phase changes, it does not
correct the response properly.
This isn't to say there aren't some cases where linear-phase changes might
not be desirable. But they're not in the majority.
A linear phase eq fails to correct the phase inequalities produced by
real-world unflatness of microphones etc. Any resonance in a mic that
might need eq will have an associated phase shift. In a "normal" eq,
this phase shift will be exactly cancelled at the same time as the
bump or dip is removed. In a linear phase one, the phase error will
remain uncorrected.
Why would you ever want that?
d
> Why would you ever want that?
The answer is... if you /weren't/ correcting an error. If you were altering
a signal that /didn't/ have response problems.
Ah - you mean a tone control. I thought we were talking about an
equaliser.
d
> Ah - you mean a tone control. I thought we were talking about
> an equaliser.
Ah, but equalizers are all-too-often used as tone controls.
If there were a Devil's Dictionary of professional recording, the definition
for "equalizer" would be: "A device often used to produce results exactly
contrary to its intended purpose."
Quite right. But the original design purpose of the equalizer was to
correct the unflatness of different microphones, so the engineer had a
well defined starting point for his operation.
d
Notching out a feedback mode on a live recording without affecting
adjacent frequencies. FIR filter is a dream come true for that.
--scott
Part of the problem is that a lot of marketing people say "linear phase"
when they really mean "minimum phase" also.
>Don Pearce <sp...@spam.com> wrote:
>>
>>A linear phase eq fails to correct the phase inequalities produced by
>>real-world unflatness of microphones etc. Any resonance in a mic that
>>might need eq will have an associated phase shift. In a "normal" eq,
>>this phase shift will be exactly cancelled at the same time as the
>>bump or dip is removed. In a linear phase one, the phase error will
>>remain uncorrected.
>>
>>Why would you ever want that?
>
>Notching out a feedback mode on a live recording without affecting
>adjacent frequencies. FIR filter is a dream come true for that.
>--scott
>
Would you call that an eq function?
>
>Part of the problem is that a lot of marketing people say "linear phase"
>when they really mean "minimum phase" also.
Totally different thing - but then I never take much notice of what
marketing people say.
d
Don,
You are such an engineer.
A normal EQ sucks and even the best audio engineer seldom achieves,
or even tries to achieve phase corrections using an EQ. Most EQs have
nothing to do with phase correction. I would be curious to better
understand what you mean by your statement quoted above.
best, eric
Of course, "linear phase" is not minimum phase. But it sounds nicer.
I mean this. Suppose a mic has - for some internal mechanical
resonance reason - a boost of 6dB at 8kHz, and the peak is 4kHz wide.
Associated with that peak is a phase shift - one way going up to the
peak, then the other way coming back down.
If you correct that peak using conventional electronic filters, such
that the peak is exactly flattened (an easy job, just specify centre
frequency and Q), then the phase shifts caused by the microphone peak
will also have been cancelled. They will no longer exist.
So you don't *try* to achieve phase correction with eq, it just
happens as an inevitable result of achieving amplitude correction.
d
It's a thing I use equalizers for all the time.
--scott
I believe a linear phase filter has the same time delay at all
frequencies. Suppose you have a 0.2 ms delay in a filter at all
frequencies - then this table would apply:
100 Hz 7.2 degrees
500 Hz 14.4 degrees
1000 Hz 36.0 degrees
2000 Hz 72.0 degrees
where phase lag is a linear function of frequency.
If phase is non-linear, then time delay is different at different
frequencies, which would seem like a bad thing.
(This is based on a recent DSP course, not on experience with audio
equipment, so FWIW.)
--
John
Good or bad isn't really the point. What you do and don't want is.
d
> If there were a Devil's Dictionary of professional recording, the definition
> for "equalizer" would be: "A device often used to produce results exactly
> contrary to its intended purpose."
I love it!
> Quite right. But the original design purpose of the equalizer was to
> correct the unflatness of different microphones, so the engineer had a
> well defined starting point for his operation.
Actually, it was to correct un-flatness in telephone lines. In studios,
it was
used to make the stuff that went into the microphone come out the way the
engineer or producer wanted to hear it. This isn't necessarily exactly like
it sounded when standing in the studio, though some say that's their goal..
>> Notching out a feedback mode on a live recording without affecting
>> adjacent frequencies. FIR filter is a dream come true for that.
>> --scott
Don Pearce wrote:
> Would you call that an eq function?
Given the common usage of the term, yes, I would. However, I wouldn't
call it "equalization." If I had to describe it, I'd call it "fixing
that annoying feedback
squeal."
>> Part of the problem is that a lot of marketing people say "linear phase"
>> when they really mean "minimum phase" also.
> Totally different thing - but then I never take much notice of what
> marketing people say.
Ah, but if everyone was like you, the marketing people wouldn't have jobs.
I know some who actually know what they're talking about and use the correct
terminology, but when writing for the masses who don't usually understand
the true meaning of the terms, you have to write something that they'll
recognize.
Leave the education to the newsgroups, forums, and what the magazines
used to be.
> I love it!
An even better definition would be "A device intended to flatten frequency
response, but most-often used to do the opposite.
> If there were a Devil's Dictionary of professional recording, the definition
> for "equalizer" would be: "A device often used to produce results exactly
> contrary to its intended purpose."
I'm sooooooo stealing that.
+1
---Jeff
> I'm sooooooo stealing that.
I wish JGH were still alive so I could share the compliments that definition
has gotten. Thank you.
> by some salesman perhaps.
Agreed.
The typical analog equalizer is minimum phase.
The usual options for DAW software and other digital equalizers are:
(1) Minimum phase equalizer.
(2) FFT-based equalizer that varies only the amplitude.
(3) FFT-based equalizer that varies only the phase
>> apparently Soundforge 5 doesn't have one - it has
>> graphic, paragraphic and parametric. And to be honest
>> I'm not altogether clear as to why these are different
>> from each other either.
A graphic equalizer is a filter bank where each filter's center frequency
spacing and andwidth are a constant fraction of an octave.
Paragraphic is not a standard term,and therefore it can be whatever
>> The paragraphic seems like 4
>> parametrics bundled together, but I assume it's more
>> than that.
>
>> What does th linear phase eq do that for example a
>> graphic eq doesn't do?
> It alters the frequency response curve without altering
> the phase response curve.
Conceptually close, but not actually true.
A linear phase filter has a phase response characteristic that is the same
as that of a simple delay.
I work with both kinds of equalizers habitually and incessantly. In the
real worlds of recording and particularly live sound equalizers are
amazingly effective tools if used properly. I feel limited unless I have at
least a 4 band full parametric equalizer or its equivalent for every channel
I'm working with. 5 or 6 bands are of course even better. You don't have to
use it all, but having it there if needed is very valuable to me.
As long as the phase shift is under a few hundred degrees and equally
applied to all relevant channels, it has minimal audible effect.
What I mean by this is that in some cases phase shift like this my be
detected in carefully controlled blind tests, but no way are you going to do
something reasonsble with it and have it pop out and stick you in the
figurative eye. The desired changes in amplitude versus frequency are far
and away the strongest audible effects.
True because so-called tone controls lack the power and flexibility to
really help in almost all real world where situations where equalization is
actually needed. Inflexible equalization is probably the most irritating
failing of cheap mixing consoles.
They can't help but do the opposite if they are used to any degree at all.
Beenthere, done that.
> Would you call that an eq function?
It often turns out that when you're notching out a frequency where feedback
is taking place, you are also reducing a response peak in the system at the
same frequency.
There is a school of thought that says that one way to equalize the response
of a live sound system is to simply turn up the gain and use your equalizer
to notch out the feedback, a number of times in succession. Crude, but not
without potential for benefit in some real-world situations.
A dirty little secret - you can have objectionable feedback in a live sound
system without there actually being any squealing.
Consider what you know about system feedback theory. Before out-and-out
oscillation, you have response peaking and then ringing. This is all
observable in live sound systems that are on the edge.
Hmm, or reverse the order and have...
Phase Linear.
The guy who did that then abused another of other technical terms, with
sonic holography coming quickly to mind.
It works, but it does leave you standing on the edge of a very steep
precipice. One little step too far and instead of a slight squeak, all
hell breaks loose.
What you are mostly finding this way is the odd peak in the
reproduction system combined with a net zero phase shift back to the
mic. That may or may not coincide with some basic room property.
Of course flatness is not a requirement of the average PA, just
loudness.
d
You need a very sharp filter to do that properly. Altec-Lansing made such a
product. It was designed by one of their engineers named Boner, and the
process was (still is?) called Boner Equalization.
Bob Carver was notorious for applying misleading names to his products
(magnetic-field amplifier, phase-coupled asymmetrical detector, etc), but
the sonic holography device did work. It might not have provided anything
remotely approaching a true hologram, but (on my own live recordings), what
came out sounded a lot more like what I heard at the mic position than what
went in.
That is a hilarious definition. Quite true.
>> There is a school of thought that says that one way to
>> equalize the response of a live sound system is to
>> simply turn up the gain and use your equalizer to notch
>> out the feedback, a number of times in succession.
>> Crude, but not without potential for benefit in some
>> real-world situations.
> You need a very sharp filter to do that properly.
It is often done with graphic equalizers with from 6 to 30 bands.
Obviously, the more bands the greater the potential for good effect, but
also the greater the reliance on the skill of the operator.
> Altec-Lansing made such a product. It was designed by one
> of their engineers named Boner, and the process was
> (still is?) called Boner Equalization.
That pretty well agrees with the history I know of which I saw evolving
before me in the trade and the literature of the 60s and 70s.
I'm under the impression that the Boners started out using custom filters
with variable Q and depth, which is what we call parametric equalization
these days.
http://digitalcontentproducer.com/mag/avinstall_equalize_let_count/
The process eventually evolved to 1/3 octave equalization which was pretty
much the industry standard for live sound and room equalization for several
decades.
One area where system equalization has been raised to a high art is
automotive sound. They work extensively with DSP parametric equalizers. They
started out with a limited number of parametric sections per signal path 3-5
years ago. In current higher end automotive sound systems the number of
available parametric sections is large. The people who do this say that
they can do a lot of good with even just two or three sections per speaker
location, but prefer having 4 or 6 available if the need presents itself.
I am pretty comfortable with 4 sections per signal path in the digital
console I use the most for live sound, given that more flexible eq is
available in the path to the main speaker system.
Or not. But that's often the case in live sound, anyway.
> One little step too far and instead
> of a slight squeak, all hell breaks loose.
Been there, done that. But you can move a mic a few feet and have that
situation, anyway.
> What you are mostly finding this way is the odd peak in
> the reproduction system combined with a net zero phase
> shift back to the mic. That may or may not coincide with
> some basic room property.
Welcome to live sound. If you have everthing right, then no problemo. But in
many cases "everything right" is unavialble.
> Of course flatness is not a requirement of the average
> PA, just loudness.
I disagree. Working at all is the first goal, but not the only one. At
least for me. I have to listen to it, so I want it to sound good!
> I disagree. Working at all is the first goal, but not the only one.
> At least for me. I have to listen to it, so I want it to sound good!
"Loudness" is directly tied to net flatness. The more resonances you remove,
the louder the PA can play.
> You need a very sharp filter to do that properly. Altec-Lansing made such a
> product. It was designed by one of their engineers named Boner, and the
> process was (still is?) called Boner Equalization.
>
>
I thought that was what happened at the b00bie bars.
/I'll be here all day, folks...
---Jeff
> As long as the phase shift is under a few hundred degrees and equally
> applied to all relevant channels, it has minimal audible effect.
Do you feel a linear phase eq is necessary to make a high quality
recording?
No. EQ, of any kind, may or may not be used.
--
ha
shut up and play your guitar
http://www.armadillomusicproductions.com/CarryMeHome.htm
http://hankalrich.com/
> > Do you feel a linear phase eq is necessary to make a high quality
> > recording?
>
> No. EQ, of any kind, may or may not be used.
Are there any circumstances you feel it's essential? I assume there
must be a reason someone came up with it.
Also, I notice in a couple of places elsewhere stated that it's not
useful for live sound because of latency - does that mean linear phase
eq is strictly a software item?
It might be possible to compensate a minimum-phase eq with an all pass
filter, but when it gets to making it adjustable it starts getting
complicated in my foggy view of this.
Scotts example of notching out a feedback howl is a great example of what
linear phase eq is good for.
Kind regards
Peter Larsen
> True because so-called tone controls lack the power and flexibility to
> really help in almost all real world where situations where
> equalization is actually needed. Inflexible equalization is probably
> the most irritating failing of cheap mixing consoles.
In my opinion it would be a great benefit to the quality of live sound if
live sound consoles had equalizers/tone controls/whatever that could only
cut.
Kind regards
Peter Larsen
J Gordon Holt, an enthusiastic amateur recordist, once observed about
non-flat microphones: "Why is it better to be 10 dB up at 10kHz than a
half-dB down?"
I can remember reading his stuff, a good writer and a good inspiration.
Kind regards
Peter Larsen
Whoever said ANY eq is necessary to make a high quality recording?
--scott
Hint: until perhaps a decade ago, FIR filters were laboratory curiosities.
Almost every recording you have ever heard was made without one.
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
> Whoever said ANY eq is necessary to make a high quality recording?
EQ starts with choosing the mic, next comes choosing the distance, so yes,
getting the EQ right is vital.
> --scott
> Hint: until perhaps a decade ago, FIR filters were laboratory
> curiosities. Almost every recording you have ever heard was made
> without one.
Hmmm ... but nnow the recordings are a LOT better, aren't they?
Kind regards
Peter Larsen
> EQ starts with choosing the mic, next comes choosing the distance,
> so yes, getting the EQ right is vital.
Non-sequitur. Choosing a microphone because it has specific colorations is
not EQ -- not in the usual sense of the word.
>> Hint: until perhaps a decade ago, FIR filters were laboratory
>> curiosities. Almost every recording you have ever heard was
>> made without one.
> Hmmm... but now the recordings are a LOT better, aren't they?
They are?
Some recordings are better because engineers are finally (again) making
recordings with minimal miking. We're finally getting recordings that can
genuinely stand comparison with live sound. And I'd be willing to bet that
EQ, of any sort, has nothing to do with their quality.
>>> Whoever said ANY EQ is necessary to make a high-quality recording?
>> EQ starts with choosing the mic, next comes choosing the distance,
>> so yes, getting the EQ right is vital.
> Non-sequitur. Choosing a microphone because it has specific
> colorations is not EQ -- not in the usual sense of the word.
>>> Hint: until perhaps a decade ago, FIR filters were laboratory
>>> curiosities. Almost every recording you have ever heard was
>>> made without one.
>> Hmmm... but nnow the recordings are a LOT better, aren't they?
> They are?
I was wondering whether to put a smiley in but thought that doubling the n
would do ........ O;-) ...
Kind regards
Peter Larsen
No. There is no "one size fits" all eq. Recordings tend to be unique, and
have unique requirements.