[Nono]

After being confused for a while after reading and hearing all this
propagating of substracting EQ-ing as a better alternative for
additive EQ-ing, I come to the conclusion that this claim is inacurate
to say the least.
It's like saying that breathing IN is better than breathing OUT.
Breathing is doing both things when appropriate.
I do recognize that it is tempting - and that happens a lot - to only
boost and totally overlook the cutting possibilities of an equalizer
and the merrits of substracting where necessary, and that more
attention should be paid to this part of the Equalizing ying/yang.
It is not a separate EQ-ing technique in itself, but a valid part of
the equalizing technique.

Some of the things that are being mentioned to claim that cutting is
better than boosting are that boosting causes phase shifts and
decreases headroom.
But to my understanding do both cutting and boosting frequencies cause
phase shifts and these phase disporportionalities are relatively as
severe on either side of the "0".
Boosting supposedly would make the phase shift more audible as the
signal becomes louder, not to mention headroom loss.
However, just like boosting would decrease headroom, would cutting
worsen the signal to noise ratio.
The answer to both the phase shift as the headroom issue is GAIN
STAGING.
If one re-adjusts the gain after EQ-ing, then a boost shouldn't affect
the phase or the headroom more severely than a cut.
For that reason I have the outputs of my recorder on the line ins
instead of the tape returns of the console, so I can adjust the gain
to always be working at the nominal operating level.

Please do not get me wrong, as I do both boosting and cutting when
mixing, while I almost only did additive EQ when I started out.
I am all for the advise to also consider the substractive aspect of
equalizing, but it is not an alternative technique. It is a just one
of the valid aspects of a technique that may or may not be better
suited to achieve the expected result, given the situation, sound or
taste.

But of course, in case I may be missing something and having it all
wrong and would welcome some enlightenment.

Regards,
Norman.

[Scott Dorsey]

Nono wrote:
After being confused for a while after reading and hearing all this
propagating of substracting EQ-ing as a better alternative for
additive EQ-ing, I come to the conclusion that this claim is inacurate
to say the least.

It is a good rule of thumb, but it is not a hard and fast rule in any
way and you don't need to overintellectualize it.

It is a good idea to think about tone shaping in terms of cutting things
out, and many equalizers sound better that way.

It is like the three-to-one rule for microphone placement; it's usually
a good idea but it's no more than that. However, it can be a good rule
to follow until you learn the exceptions.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Nono]

On 16 jul, 22:34, (Scott Dorsey) wrote:
Nono wrote:
After being confused for a while after reading and hearing all this
propagating of substracting EQ-ing as a better alternative for
additive EQ-ing, I come to the conclusion that this claim is inacurate
to say the least.

It is a good rule of thumb, but it is not a hard and fast rule in any
way and you don't need to overintellectualize it.


Yeah, I tend to do that, LOL.

[Mike Rivers]

Nono wrote:
After being confused for a while after reading and hearing all this
propagating of substracting EQ-ing as a better alternative for
additive EQ-ing, I come to the conclusion that this claim is inacurate
to say the least.

People write all kinds of crap on the 'net, and readers
interpret it in different ways. I think that the concept of
trying to fix things by cutting rather than boosting is a
good starting place for a couple of reasons, but it depends
on why you're using the EQ in the first place.

When equalizing a microphone signal, cutting is often the
preferred first choice because the peaks in a mic's response
are often more bothersome than the dips. Generally things
sound better when taking out the peakiness than when you try
to fill in the gaps between the peaks.

There are two primary reasons for equalizing a sound system.
One is to reduce feedback. Feedback is caused by too much
gain at a specific frequency, so cutting that frequency
often helps. The other reason is to smooth the transition
between speakers in a bi- or tri-amped system. While it's
beter to try to smooth that out by the proper choice of
crossover frequency, more often than not you end up with a
hump. Again, reducing the hump is better than trying to make
what's on either side as loud as the hump.

Using EQ to make things fit into a mix is the classic
example of cut first. Generally you use EQ in this instance
because two instruments are (because of what they are) have
sizeable components of their sound in the same frequency
range. By cutting that range in one or both instruments,
they are easier to distinguish in the mix. They no longer
sound like what went into the mic, but the mix benefits. If
you wanted to not have to do this sort of buggery, you would
have orchestrated the music so that the instruments didn't
compete for the same portion of the spectrum.

When using EQ to create a sound that an instrument or other
source didn't produce, anything goes. Whatever gets you the
results is OK, but you'll rarely find that you can get the
same result both by boosting or cutting, simply because of
the characteristics of filters.

Some of the things that are being mentioned to claim that cutting is
better than boosting are that boosting causes phase shifts and
decreases headroom.

Boosting raises the signal level, at least at the boosted
frequency. If you don't lower the overall volume after
equalizing, you'll have moved closer to the maximum level of
the system, which means that you've reduced your headroom.
This is something that's under operator control of course,
but if you boost a frequency range to bring it out in the
mix, then turn down the level to keep your headroom, you may
still not be able to hear what you wanted to hear. Perhaps a
better approach is to turn down (or cut-EQ) something else.

Everything causes phase shift. That's what makes an
equalizer work. This is a bad thing?

Boosting supposedly would make the phase shift more audible as the
signal becomes louder, not to mention headroom loss.

We don't "hear" this sort of phase shift, other than in the
manner that it affects the frequency response. And isn't
that what we're after?

However, just like boosting would decrease headroom, would cutting
worsen the signal to noise ratio.

Not necessarily, in fact it can often improve it if what
you're cutting is considered noise. But if you have to boost
a track after cutting something with EQ, you could increase
the noise. However if this is where you end up, you're
probably adjusting the wrong thing.

The answer to both the phase shift as the headroom issue is GAIN
STAGING.

Sorry, I thought you were asking a question. I didn't
realize you were about to give us a lecture.

If one re-adjusts the gain after EQ-ing, then a boost shouldn't affect
the phase or the headroom more severely than a cut.
For that reason I have the outputs of my recorder on the line ins
instead of the tape returns of the console, so I can adjust the gain
to always be working at the nominal operating level.

Maybe I'm misunderstanding. What recorder? What console?
Don't you have gain controls?

I almost only did additive EQ when I started out.
I am all for the advise to also consider the substractive aspect of
equalizing, but it is not an alternative technique. It is a just one
of the valid aspects of a technique that may or may not be better
suited to achieve the expected result, given the situation, sound or
taste.

Exactly.
..


--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

[Nono]

On 16 jul, 23:56, Mike Rivers wrote:
Nono wrote:
After being confused for a while after reading and hearing all this
propagating of substracting EQ-ing as a better alternative for
additive EQ-ing, I come to the conclusion that this claim is inacurate
to say the least.

People write all kinds of crap on the 'net, and readers
interpret it in different ways. I think that the concept of
trying to fix things by cutting rather than boosting is a
good starting place for a couple of reasons, but it depends
on why you're using the EQ in the first place.

When equalizing a microphone signal, cutting is often the
preferred first choice because the peaks in a mic's response
are often more bothersome than the dips. Generally things
sound better when taking out the peakiness than when you try
to fill in the gaps between the peaks.

There are two primary reasons for equalizing a sound system.
One is to reduce feedback. Feedback is caused by too much
gain at a specific frequency, so cutting that frequency
often helps. The other reason is to smooth the transition
between speakers in a bi- or tri-amped system. While it's
beter to try to smooth that out by the proper choice of
crossover frequency, more often than not you end up with a
hump. Again, reducing the hump is better than trying to make
what's on either side as loud as the hump.

Using EQ to make things fit into a mix is the classic
example of cut first. Generally you use EQ in this instance
because two instruments are (because of what they are) have
sizeable components of their sound in the same frequency
range. By cutting that range in one or both instruments,
they are easier to distinguish in the mix. They no longer
sound like what went into the mic, but the mix benefits. If
you wanted to not have to do this sort of buggery, you would
have orchestrated the music so that the instruments didn't
compete for the same portion of the spectrum.

When using EQ to create a sound that an instrument or other
source didn't produce, anything goes. Whatever gets you the
results is OK, but you'll rarely find that you can get the
same result both by boosting or cutting, simply because of
the characteristics of filters.

Some of the things that are being mentioned to claim that cutting is
better than boosting are that boosting causes phase shifts and
decreases headroom.

Boosting raises the signal level, at least at the boosted
frequency. If you don't lower the overall volume after
equalizing, you'll have moved closer to the maximum level of
the system, which means that you've reduced your headroom.
This is something that's under operator control of course,
but if you boost a frequency range to bring it out in the
mix, then turn down the level to keep your headroom, you may
still not be able to hear what you wanted to hear. Perhaps a
better approach is to turn down (or cut-EQ) something else.

Everything causes phase shift. That's what makes an
equalizer work. This is a bad thing?

Boosting supposedly would make the phase shift more audible as the
signal becomes louder, not to mention headroom loss.

We don't "hear" this sort of phase shift, other than in the
manner that it affects the frequency response. And isn't
that what we're after?

However, just like boosting would decrease headroom, would cutting
worsen the signal to noise ratio.

Not necessarily, in fact it can often improve it if what
you're cutting is considered noise. But if you have to boost
a track after cutting something with EQ, you could increase
the noise. However if this is where you end up, you're
probably adjusting the wrong thing.

The answer to both the phase shift as the headroom issue is GAIN
STAGING.

Sorry, I thought you were asking a question. I didn't
realize you were about to give us a lecture.

If one re-adjusts the gain after EQ-ing, then a boost shouldn't affect
the phase or the headroom more severely than a cut.
For that reason I have the outputs of my recorder on the line ins
instead of the tape returns of the console, so I can adjust the gain
to always be working at the nominal operating level.

Maybe I'm misunderstanding. What recorder? What console?
Don't you have gain controls?

I almost only did additive EQ when I started out.
I am all for the advise to also consider the substractive aspect of
equalizing, but it is not an alternative technique. It is a just one
of the valid aspects of a technique that may or may not be better
suited to achieve the expected result, given the situation, sound or
taste.

Exactly.
.

--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson


Your post already gave more insight than most of the articles that I
read or misunderstood on the subject.
I wasn't asking a question, but I surely wasn't trying to give any
lecture to anybody, so if my statement about the GAIN STAGING was
wrong, I would have appreciated it more if you just told me that
instead of acusing me of giving lectures.
But I do know my place now and I'll just crawl back to my corner,
leaving the floor to the grown ups.

Good bye.

[Mike Rivers]

Nono wrote:

Your post already gave more insight than most of the articles that I
read or misunderstood on the subject.
I wasn't asking a question, but I surely wasn't trying to give any
lecture to anybody, so if my statement about the GAIN STAGING was
wrong, I would have appreciated it more if you just told me that
instead of acusing me of giving lectures.
But I do know my place now and I'll just crawl back to my corner,
leaving the floor to the grown ups.

Why do they always take things the wrong way?

Your statement about gain staging was reasonably accurate
(otherwise I would have corrected you), but it sounded like
you were presenting your "A-HA! moment" as if this was the
solution to the headroom problem that was caused by
EQ-by-boosting.

--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

[Ethan Winer[_3_]]

Great post Norman. Mike Rivers gave an excellent detailed reply, so
I'll address only this part:

On Jul 16, 4:26 pm, Nono wrote:
Some of the things that are being mentioned to claim that cutting is
better than boosting are that boosting causes phase shifts and
decreases headroom.

Usually I see people claim that phase shift itself is a problem, which
is not true. Phase shift in the typical amounts you get from a normal
EQ are not only benign, but a necessary part of the process. I address
this in my EQ and Phase shift article:

http://www.ethanwiner.com/EQPhase.html

More here by David Clark:

http://www.ethanwiner.com/phase.html

You can hear audio examples of phase shift in my AES video and support
files:

http://www.youtube.com/watch?v=BYTlN6wjcvQ
http://www.ethanwiner.com/aes/

As for phase shift reducing headroom, it may or it may not, depending
on the original signal. Not that 1 dB of headroom is so precious
anyway. The Orban Optimod includes a Phase Rotator that claims to
increase headroom letting broadcasters achieve higher signal levels on
voices before the onset of clipping:

http://www.orban.com/products/radio/...0/signal_flow/

--Ethan

[Roy W. Rising[_2_]]

Boost can compromise headroom in a poorly designed signal chain.

*Any* analog EQ with a slope exceeding 6dB/octave introduces irreversible
and sometimes noticeable-to-unacceptable phase shifts.

I wish I'd known the latter earlier in my career.

--
~ Roy
"If you notice the sound, it's wrong!"

[Les Cargill[_3_]]

Roy W. Rising wrote:
Boost can compromise headroom in a poorly designed signal chain.

*Any* analog EQ with a slope exceeding 6dB/octave introduces irreversible
and sometimes noticeable-to-unacceptable phase shifts.


Phase shift is the point of EQ. It don't hurt. Pan the original and
post-EQ left and right - they don't sound 'phasey'.

The problems of EQ abuse generally have more to do with headroom .

I wish I'd known the latter earlier in my career.


--
Les Cargill

[William Sommerwerck]

*Any* analog EQ with a slope exceeding 6dB/octave introduces
irreversible and sometimes noticeable-to-unacceptable phase shifts.

This is simply not true. You need to do some reading up on the mathematics
of transfer functions.

If I use minimum-phase EQ to exactly cancel out a minimum-phase error, then
the amplitude response is flat, and there is no net phase shift. It doesn't
matter what the slopes were. And, /any/ EQ is, in principle, reversible. If
it's not minimum phase, you have to know (or figure out) what it was.

Digital EQ is not /inherently/ different form analog EQ. It does, however,
permit a much wider range of control over phase shift.

[Don Pearce[_3_]]

On 17 Jul 2010 17:45:23 GMT, Roy W. Rising
wrote:

"William Sommerwerck" wrote:
*Any* analog EQ with a slope exceeding 6dB/octave introduces
irreversible and sometimes noticeable-to-unacceptable phase shifts.

This is simply not true. You need to do some reading up on the
mathematics of transfer functions.

If I use minimum-phase EQ to exactly cancel out a minimum-phase error,
then the amplitude response is flat, and there is no net phase shift. It
doesn't matter what the slopes were. And, /any/ EQ is, in principle,
reversible. If it's not minimum phase, you have to know (or figure out)
what it was.

Digital EQ is not /inherently/ different form analog EQ. It does,
however, permit a much wider range of control over phase shift.

If it didn't matter, "minimum phase" would not be of interest.

What do you believe "minimum phase" actually means?

d

[Roy W. Rising[_2_]]

Les Cargill wrote:
Roy W. Rising wrote:
Boost can compromise headroom in a poorly designed signal chain.

*Any* analog EQ with a slope exceeding 6dB/octave introduces
irreversible and sometimes noticeable-to-unacceptable phase shifts.


Phase shift is the point of EQ. It don't hurt. Pan the original and
post-EQ left and right - they don't sound 'phasey'.

The problems of EQ abuse generally have more to do with headroom .

I wish I'd known the latter earlier in my career.

I have to disagree. Amplitude adjustment is the point of EQ. When the
harmonics of a note are shifted by a slope greater than 6dB/octave, it
ceases to sound "right". Really steep slopes cause objectionable phase
distortions. I've heard them!

--
~ Roy
"If you notice the sound, it's wrong!"

[Roy W. Rising[_2_]]

"William Sommerwerck" wrote:
*Any* analog EQ with a slope exceeding 6dB/octave introduces
irreversible and sometimes noticeable-to-unacceptable phase shifts.

This is simply not true. You need to do some reading up on the
mathematics of transfer functions.

If I use minimum-phase EQ to exactly cancel out a minimum-phase error,
then the amplitude response is flat, and there is no net phase shift. It
doesn't matter what the slopes were. And, /any/ EQ is, in principle,
reversible. If it's not minimum phase, you have to know (or figure out)
what it was.

Digital EQ is not /inherently/ different form analog EQ. It does,
however, permit a much wider range of control over phase shift.

If it didn't matter, "minimum phase" would not be of interest.

--
~ Roy
"If you notice the sound, it's wrong!"

[William Sommerwerck]

Part of the issue is the meaning of EQ. If Ambrose Bierce were writing a
dictionary of professional audio terms, we would likely read...

equalizer: A device used to produce exactly the opposite effect implied
by its name.

When "equalizing" tracks, an engineer is almost always introducing what is,
objectively, an error into the signal.

[Don Pearce[_3_]]

On Sat, 17 Jul 2010 10:55:38 -0700, "William Sommerwerck"
wrote:

Part of the issue is the meaning of EQ. If Ambrose Bierce were writing a
dictionary of professional audio terms, we would likely read...

equalizer: A device used to produce exactly the opposite effect implied
by its name.

When "equalizing" tracks, an engineer is almost always introducing what is,
objectively, an error into the signal.


For most engineers an equaliser is nothing more than a fancy tone
control. Its original purpose was to make all incoming signals from
whatever source nominally flat, so you had a known starting point.

d

[Mike Rivers]

Ethan Winer wrote:

As for phase shift reducing headroom, it may or it may not, depending
on the original signal. Not that 1 dB of headroom is so precious
anyway. The Orban Optimod includes a Phase Rotator that claims to
increase headroom letting broadcasters achieve higher signal levels on
voices before the onset of clipping:

"Phase rotator" is kind of a misnomer. Normally, harmonics
are in phase with the fundamental and therefore sum to a
greater amplitude than the fundamental itself. This kind of
phase rotator delays the harmonics by different amounts so
that the peaks of the harmonics don't sum with the peaks of
the fundamental. It's practically impossible to notice on
speech, but it can make a noticeable change in the timbre of
certain instruments.


--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

[Scott Dorsey]

Roy W. Rising wrote:
Boost can compromise headroom in a poorly designed signal chain.

Yes.

*Any* analog EQ with a slope exceeding 6dB/octave introduces irreversible
and sometimes noticeable-to-unacceptable phase shifts.

They aren't irreversable! They can be removed by using the same EQ in
reverse.

In a minimum phase system (and most things you will encounter in the audio
world are minimum phase), the phase and amplitude response with frequency
go hand in and and you can calculate one from the other.

This means that if you have a response problem in a mike, it will cause
both amplitude and phase effects, and that if you could fix one with EQ
the other one gets fixed along with it, for free.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Scott Dorsey]

Roy W. Rising wrote:
I have to disagree. Amplitude adjustment is the point of EQ. When the
harmonics of a note are shifted by a slope greater than 6dB/octave, it
ceases to sound "right". Really steep slopes cause objectionable phase
distortions. I've heard them!

I bet a nickel what you are hearing isn't the result of phase shift.

I'd bet a penny that it's the result of harmonic distortion in the
equalizer, which is of course inherent in all electronics. And that's
a good reason to avoid aggressive EQ, as much as any.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Adrian Tuddenham[_2_]]

William Sommerwerck wrote:

Part of the issue is the meaning of EQ. If Ambrose Bierce were writing a
dictionary of professional audio terms, we would likely read...

equalizer: A device used to produce exactly the opposite effect implied
by its name.

When "equalizing" tracks, an engineer is almost always introducing what is,
objectively, an error into the signal.

I have objected for may years to the mis-application of the term
'equaliser'. An equaliser objectively corrects a known frequency
response distortion. The distortion is is often the result of an
earlier deliberate and carefully-controlled process.

A graphic effects unit almost never does that - it is not an equaliser
and should not be called one.

--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk

[Roy W. Rising[_2_]]

(Don Pearce) wrote:
On 17 Jul 2010 17:45:23 GMT, Roy W. Rising
wrote:

"William Sommerwerck" wrote:
*Any* analog EQ with a slope exceeding 6dB/octave introduces
irreversible and sometimes noticeable-to-unacceptable phase shifts.

This is simply not true. You need to do some reading up on the
mathematics of transfer functions.

If I use minimum-phase EQ to exactly cancel out a minimum-phase error,
then the amplitude response is flat, and there is no net phase shift.
It doesn't matter what the slopes were. And, /any/ EQ is, in
principle, reversible. If it's not minimum phase, you have to know (or
figure out) what it was.

Digital EQ is not /inherently/ different form analog EQ. It does,
however, permit a much wider range of control over phase shift.

If it didn't matter, "minimum phase" would not be of interest.

What do you believe "minimum phase" actually means?

d
To me, "zero phase" means the phase relationships of the harmonic structure
of a sound are not changed by the boost or cut of amplitude within the
frequency range of the sound. "Minimum phase" means an effort to reach
"zero phase" has been made.

--
~ Roy
"If you notice the sound, it's wrong!"

[Don Pearce[_3_]]

On 17 Jul 2010 18:47:41 GMT, Roy W. Rising
wrote:

(Don Pearce) wrote:
On 17 Jul 2010 17:45:23 GMT, Roy W. Rising
wrote:

"William Sommerwerck" wrote:
*Any* analog EQ with a slope exceeding 6dB/octave introduces
irreversible and sometimes noticeable-to-unacceptable phase shifts.

This is simply not true. You need to do some reading up on the
mathematics of transfer functions.

If I use minimum-phase EQ to exactly cancel out a minimum-phase error,
then the amplitude response is flat, and there is no net phase shift.
It doesn't matter what the slopes were. And, /any/ EQ is, in
principle, reversible. If it's not minimum phase, you have to know (or
figure out) what it was.

Digital EQ is not /inherently/ different form analog EQ. It does,
however, permit a much wider range of control over phase shift.

If it didn't matter, "minimum phase" would not be of interest.

What do you believe "minimum phase" actually means?

d
To me, "zero phase" means the phase relationships of the harmonic structure
of a sound are not changed by the boost or cut of amplitude within the
frequency range of the sound. "Minimum phase" means an effort to reach
"zero phase" has been made.

I thought you might be picturing something like that. No, it has
nothing to do with any of that. It is a purely technical term
concerning the stability (among other things) of the system, and
particularly the location of poles and circles within the unit circle.

Now - to your thoughts. Natural frequency response errors are caused
by frequency-dependent transfer functions. This could be a room
resonance, the mass of a mic diaphragm or a length of mismatched
cable. All of these errors are accompanied by a phase shift which is
entirely predictable since these are all minimum phase changes. What
this means is that they are totally correctable in both amplitude and
phase even if you can only measure the amplitude error. So an
equaliser, if you want it to do its job, MUST exhibit the phase
changes associated with the frequency response. If it doesn't, the
signal won't be corrected and phase errors will remain.

d

[Don Pearce[_3_]]

On Sat, 17 Jul 2010 19:01:43 GMT, (Don Pearce) wrote:

On 17 Jul 2010 18:47:41 GMT, Roy W. Rising
wrote:

(Don Pearce) wrote:
On 17 Jul 2010 17:45:23 GMT, Roy W. Rising
wrote:

"William Sommerwerck" wrote:
*Any* analog EQ with a slope exceeding 6dB/octave introduces
irreversible and sometimes noticeable-to-unacceptable phase shifts.

This is simply not true. You need to do some reading up on the
mathematics of transfer functions.

If I use minimum-phase EQ to exactly cancel out a minimum-phase error,
then the amplitude response is flat, and there is no net phase shift.
It doesn't matter what the slopes were. And, /any/ EQ is, in
principle, reversible. If it's not minimum phase, you have to know (or
figure out) what it was.

Digital EQ is not /inherently/ different form analog EQ. It does,
however, permit a much wider range of control over phase shift.

If it didn't matter, "minimum phase" would not be of interest.

What do you believe "minimum phase" actually means?

d
To me, "zero phase" means the phase relationships of the harmonic structure
of a sound are not changed by the boost or cut of amplitude within the
frequency range of the sound. "Minimum phase" means an effort to reach
"zero phase" has been made.

I thought you might be picturing something like that. No, it has
nothing to do with any of that. It is a purely technical term
concerning the stability (among other things) of the system, and
particularly the location of poles and circles within the unit circle.


Sorry about that. Poles and zeroes within the unit circle.

Now - to your thoughts. Natural frequency response errors are caused
by frequency-dependent transfer functions. This could be a room
resonance, the mass of a mic diaphragm or a length of mismatched
cable. All of these errors are accompanied by a phase shift which is
entirely predictable since these are all minimum phase changes. What
this means is that they are totally correctable in both amplitude and
phase even if you can only measure the amplitude error. So an
equaliser, if you want it to do its job, MUST exhibit the phase
changes associated with the frequency response. If it doesn't, the
signal won't be corrected and phase errors will remain.

d

[William Sommerwerck]

What do you believe "minimum phase" actually means?

To me, "zero phase" means the phase relationships of the harmonic
structure of a sound are not changed by the boost or cut of amplitude
within the frequency range of the sound. "Minimum phase" means
an effort to reach "zero phase" has been made.

That's common sense, isn't it? But it's incorrect.

A minimum-phase filter has a phase characteristic that is directly related
to its frequency response by the Hilbert transform. The phase shift is the
lowest possible phase shift for that particular amplitude transfer
characteristic, hence the name.

There's more to it than that. I can't think at the moment of a book to
recommend.

[William Sommerwerck]

Now - to your thoughts. Natural frequency response errors are caused
by frequency-dependent transfer functions. This could be a room
resonance, the mass of a mic diaphragm or a length of mismatched
cable. All of these errors are accompanied by a phase shift which is
entirely predictable since these are all minimum phase changes. What
this means is that they are totally correctable in both amplitude and
phase even if you can only measure the amplitude error. So an
equaliser, if you want it to do its job, MUST exhibit the phase
changes associated with the frequency response. If it doesn't, the
signal won't be corrected and phase errors will remain.

correct

[Les Cargill[_3_]]

Roy W. Rising wrote:
Les wrote:
Roy W. Rising wrote:
Boost can compromise headroom in a poorly designed signal chain.

*Any* analog EQ with a slope exceeding 6dB/octave introduces
irreversible and sometimes noticeable-to-unacceptable phase shifts.


Phase shift is the point of EQ. It don't hurt. Pan the original and
post-EQ left and right - they don't sound 'phasey'.

The problems of EQ abuse generally have more to do with headroom .

I wish I'd known the latter earlier in my career.

I have to disagree. Amplitude adjustment is the point of EQ.

From a Fourier perspective, sure. From a "how do RC filters work"
it's all about the time constant.

When the
harmonics of a note are shifted by a slope greater than 6dB/octave, it
ceases to sound "right". Really steep slopes cause objectionable phase
distortions. I've heard them!


I've heard non-zero-group-delay EQ in the digital domain with
-24dB/octave slope and I *don't* hear it (in the passband).
YMMV.

--
Les Cargill

[Scott Dorsey]

William Sommerwerck wrote:

A minimum-phase filter has a phase characteristic that is directly related
to its frequency response by the Hilbert transform. The phase shift is the
lowest possible phase shift for that particular amplitude transfer
characteristic, hence the name.

There's more to it than that. I can't think at the moment of a book to
recommend.

Audio Cyclopedia is as good as any, I think.

But Cadzow's _Signals, Systems, and Transforms_ is probably more up to date.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[William Sommerwerck]

I've heard non-zero-group-delay EQ in the digital domain with
-24dB/octave slope and I *don't* hear it (in the passband).

If the filter is truly /constant/ group delay, then you should hear no
difference, other than the intended FR alteration.

[Arny Krueger]

"Roy W. Rising" wrote in
message

*Any* analog EQ with a slope exceeding 6dB/octave
introduces irreversible and sometimes
noticeable-to-unacceptable phase shifts.

Seems like you forgot quite a bit that they tried to teach us in our EE
classes, or perhaps you never learned?

Cascade two differentiatiors followed by two integrators. At the output of
the second differentiator, the frequency response is +12 dB/octave and there
is strongly leading phase. At the output of the second integrator, the
response is flat and the phase is flat. Use high quality circuits and
reasonble levels and there will be no audible problems with noise or
distortion. Your poorly thought out assertion is disproven.

[Arny Krueger]

"William Sommerwerck" wrote in
message
What do you believe "minimum phase" actually means?

To me, "zero phase" means the phase relationships of the
harmonic structure of a sound are not changed by the
boost or cut of amplitude within the frequency range of
the sound. "Minimum phase" means
an effort to reach "zero phase" has been made.

That's common sense, isn't it? But it's incorrect.

A minimum-phase filter has a phase characteristic that is
directly related to its frequency response by the Hilbert
transform. The phase shift is the lowest possible phase
shift for that particular amplitude transfer
characteristic, hence the name.

There's more to it than that. I can't think at the moment
of a book to recommend.

Just about any book that used to be used in second year EE courses.

I was waiting for someone to give the right answer. I guess that makes 3 of
us who do know, and apparently a raft of people that don't.

[Ty Ford]

On Fri, 16 Jul 2010 16:26:12 -0400, Nono wrote
(in article
):

Please do not get me wrong, as I do both boosting and cutting when mixing,
while I almost only did additive EQ when I started out. I am all for the
advise to also consider the substractive aspect of equalizing,

Then why bother us with all the back and forth?

Regards,

Ty Ford

--Audio Equipment Reviews Audio Production Services
Acting and Voiceover Demos http://www.tyford.com
Guitar player?:http://www.youtube.com/watch?v=yWaPRHMGhGA

[Ethan Winer[_3_]]

On Jul 17, 1:43 pm, Roy W. Rising
wrote:
Really steep slopes cause objectionable phase distortions. I've heard them!

How do you know that what you heard was phase shift? Using the best
descriptive words you can muster, describe what you believe to be the
objectionable part of a roll-off with a slope steeper than 6 dB per
octave.

--Ethan

[Ethan Winer[_3_]]

On Jul 17, 2:38 pm, Mike Rivers wrote:
"Phase rotator" is kind of a misnomer. Normally, harmonics
are in phase with the fundamental and therefore sum to a
greater amplitude than the fundamental itself. This kind of
phase rotator delays the harmonics by different amounts so
that the peaks of the harmonics don't sum with the peaks of
the fundamental. It's practically impossible to notice on
speech, but it can make a noticeable change in the timbre of
certain instruments.

Interesting Mike. I always ASSumed the Orban device was a simple all-
pass filter. I've seen an all-pass filter change the height of peaks
in a waveform. But you're saying what the Orban does is more
elaborate?

--Ethan

[William Sommerwerck]

Really steep slopes cause objectionable phase distortions.
I've heard them!

If you're applying the "EQ" to a signal that hasn't been processed and is
"flat" to begin with, this might very be true.

But if you're trying to fix an error in a mic or a room, the EQ should make
things better.

[Ty Ford]

On Mon, 19 Jul 2010 12:43:31 -0400, Ethan Winer wrote
(in article
):

On Jul 17, 2:38 pm, Mike Rivers wrote:
"Phase rotator" is kind of a misnomer. Normally, harmonics
are in phase with the fundamental and therefore sum to a
greater amplitude than the fundamental itself. This kind of
phase rotator delays the harmonics by different amounts so
that the peaks of the harmonics don't sum with the peaks of
the fundamental. It's practically impossible to notice on
speech, but it can make a noticeable change in the timbre of
certain instruments.

Interesting Mike. I always ASSumed the Orban device was a simple all-
pass filter. I've seen an all-pass filter change the height of peaks
in a waveform. But you're saying what the Orban does is more
elaborate?

--Ethan

When speaking in real-time though one of Bob's boxes with the phase rotators
engaged and headphones on, your own voice sounds "stuffy", sort of like you
have a head cold.

Regards,

Ty Ford

I think the idea of phase rotation to equalize positive and negative peaks
was a useful scheme for AM transmitters, where peak modulation was limited to
+125% (by the FCC, back in the 1970's, I forget the year).

Before that, it was not at all unusual to see plate modulated AM transmitter
with 185% positive peaks. Since output power is a function of modulation in
AM, evening and packing the modulation envelope resulted in firmer reception
on the fringes.

Most voices are asymmetrical to some degree. Mine peaks more negatively than
positively. Sometimes that actually results in a voice that cuts through more
(for better or worse) on a recorded track or mix.

Regards,

Ty Ford



--Audio Equipment Reviews Audio Production Services
Acting and Voiceover Demos http://www.tyford.com
Guitar player?:http://www.youtube.com/watch?v=yWaPRHMGhGA

[William Sommerwerck]

"Phase rotator" is kind of a misnomer. Normally, harmonics
are in phase with the fundamental and therefore sum to a
greater amplitude than the fundamental itself. This kind of
phase rotator delays the harmonics by different amounts so
that the peaks of the harmonics don't sum with the peaks
of the fundamental. It's practically impossible to notice on
speech, but it can make a noticeable change in the timbre
of certain instruments.

Interesting Mike. I always ASSumed the Orban device was a
simple all-pass filter. I've seen an all-pass filter change the height
of peaks in a waveform. But you're saying what the Orban does
is more elaborate?

What do you mean by a "simple" all-pass filter? The simplest kind -- such as
the one found at the input of most oscilloscopes -- has flat frequency
response and no net phase shift. It's "useless", other than to "null out"
the effects of cable capacitance. *

I'm not sure why you'd want to use an all-pass filter that altered phase
relationships, unless you were correcting for some known error, such as
non-linear group delay (eg, analog tape recording).

No one appears to have incorporated such a filter in consumer digital
control centers. It would be nice to be able to /advance/ the phase of
higher frequencies without altering the amplitude response.


* It also attenuates the input. The two unavoidably go together.

[Scott Dorsey]

William Sommerwerck wrote:
I'm not sure why you'd want to use an all-pass filter that altered phase
relationships, unless you were correcting for some known error, such as
non-linear group delay (eg, analog tape recording).

The goal is to make sure you have equal number of positive-going and
negative-going peaks. You throw group delay in and out of the circuit
constantly in such a way that your waveform becomes symmetric. This
increases your channel utilization without changing the perceived sound much.

It is one weapon in the armory of loudness-increasing devices.

No one appears to have incorporated such a filter in consumer digital
control centers. It would be nice to be able to /advance/ the phase of
higher frequencies without altering the amplitude response.

You would want an IBP from Little Labs for that. It is a handy device.
--scott


--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[William Sommerwerck]

You would want an IBP from Little Labs for that.

It's a step in the right direction, but not exactly what I'm thinking of.

[Steven Borrelli]

On Jul 16, 3:26*pm, Nono wrote:
After being confused for a while after reading and hearing all this
propagating of substracting EQ-ing as a better alternative for
additive EQ-ing, I come to the conclusion that this claim is inacurate
to say the least.
It's like saying that breathing IN is better than breathing OUT.
Breathing is doing both things when appropriate.
I do recognize that it is tempting - and that happens a lot - to only
boost and totally overlook the cutting possibilities of an equalizer
and the merrits of substracting where necessary, and that more
attention should be paid to this part of the Equalizing ying/yang.
It is not a separate EQ-ing technique in itself, but a valid part of
the equalizing technique.

Some of the things that are being mentioned to claim that cutting is
better than boosting are that boosting causes phase shifts and
decreases headroom.
But to my understanding do both cutting and boosting frequencies cause
phase shifts and these phase disporportionalities are relatively as
severe on either side of the "0".
Boosting supposedly would make the phase shift more audible as the
signal becomes louder, not to mention headroom loss.
However, just like boosting would decrease headroom, would cutting
worsen the signal to noise ratio.
The answer to both the phase shift as the headroom issue is GAIN
STAGING.
If one re-adjusts the gain after EQ-ing, then a boost shouldn't affect
the phase or the headroom more severely than a cut.
For that reason I have the outputs of my recorder on the line ins
instead of the tape returns of the console, so I can adjust the gain
to always be working at the nominal operating level.

Please do not get me wrong, as I do both boosting and cutting when
mixing, while I almost only did additive EQ when I started out.
I am all for the advise to also consider the substractive aspect of
equalizing, but it is not an alternative technique. It is a just one
of the valid aspects of a technique that may or may not be better
suited to achieve the expected result, given the situation, sound or
taste.

But of course, in case I may be missing something and having it all
wrong and would welcome some enlightenment.

Regards,
Norman.

Just follow your ears.

[Roy W. Rising[_2_]]

Ethan Winer wrote:
On Jul 17, 1:43 pm, Roy W. Rising
wrote:
Really steep slopes cause objectionable phase distortions. I've heard
them!

How do you know that what you heard was phase shift? Using the best
descriptive words you can muster, describe what you believe to be the
objectionable part of a roll-off with a slope steeper than 6 dB per
octave.

--Ethan

OK. I do not know that the cause was phase shift. That said, the mode was
not roll-off. The device was a '60's McCurdy EQ with 3- or 6KHz haystack
boost. When used with more than about 6dB of 6KHz boost, sources had what
I can describe only as a "pinched" characteristic. I think it was masked
in the JBL monitors being used, it became evident when EV Sentry 500s
became available. The system was very well engineered, I seriously doubt
anything down stream was being overloaded.

In another circumstance, a UREI 1/3 octave graphic EQ was being used to
correct for known minor discrepancies in a 'standard' mic. A group of
'golden ears' listened to reference material and characterized what they
heard as "something wrong". None could be any clearer. When the EQ was
flattened, all agreed to leave it that way. Whatever was bothering them
was gone.

I can't remember which publication long ago raised this topic and offered
the "6dB/octave" suggestions. The arguments were persuasive.

--
~ Roy
"If you notice the sound, it's wrong!"

[Roy W. Rising[_2_]]

"William Sommerwerck" wrote:
Really steep slopes cause objectionable phase distortions.
I've heard them!

If you're applying the "EQ" to a signal that hasn't been processed and is
"flat" to begin with, this might very be true.

Indeed, this was the case.

But if you're trying to fix an error in a mic or a room, the EQ should
make things better.

That's what we've been led to believe. I'm not fully convinced.

--
~ Roy
"If you notice the sound, it's wrong!"