Old 23rd February 2013
  #1
Gear addict
 

Thread Starter
Phase rotator?

I noticed that Izotope RX2 has a "phase rotator" tool. (Different than a phase "shift" tool). It's purpose is to make asymmetrical waveforms more symmetric - and gains a few dB of level by doing so.

According to a paper I read by Bob Orban, radio broadcasting processors use phase rotators to squeeze a few more dB out of audio program - mostly dialog.

If you used a phase rotator on music or voice during mixing or mastering wouldn't that negate the effect, or create some other weird phase effect, if that recording was subsequently broadcast?

What would this tool be used for?
Old 23rd February 2013
  #2
Lives for gear
 
wado1942's Avatar
 

I thought about this. Only certain types of sounds benefit from this process at the cost of a slight loss in clarity. If you apply, say 90 degrees of phase shift to the signal, you can get 1-2dB lower peak to average ratio. No phasing effects as you'd recognize would be heard because you're hearing ONLY the phase rotated signal and not the dry signal. Another 90 degree phase shift added to the signal in the radio processor would equal a 180 degree phase shift, reducing any advantage you got in the first place.
Honestly, I think phase rotators can go away in radio processors because they wreak havoc with clipped audio (read, all the material getting fed into them these days) and non-clipped, dialogue-based signals are now notably louder than the main program material.
Old 24th February 2013
  #3
I've asked Bill and Kim Sacks (Orban engineers) the same question and your assumptions are correct, I agree with wado1942 as well. I have had one rare exception when a + or -90 degree phase shift worked out.

Without going into another long winded discussion on polarity and if we can hear it or not, In my room I can. My speaker outputs are polarity correct, I know I can't control what happens in the real world, but putting things straight makes me feel good, nuff said (I hope!!)

Sometimes I'll get something that though it seems to have plenty of lo-end, still sounds flat. Reversing the polarity makes no difference, but in some rare (very rare) cases pushing the polarity 90 deg in one direction or the other brings the life back into the lo-end. It's as if eq was used near the fundamental of the kick, distorting the phase response in a negative way?? Guessing here, those who are smarter than I or have more free time please chime in. As far as I can tell there are rare cases that a 90 degree phase shift creates Asymmetry that puts things as they should be.

Curious what others think.
Old 25th February 2013
  #4
Lives for gear
 
echoRausch's Avatar
 

Maybe Aleksej should chime in here.

As far as I know RX2 has an algorithm to calculate the "perfect" amount of phase shift to optimize the signal. The result is usually some more headroom.
Old 25th February 2013
  #5
Lives for gear
 
huejahfink's Avatar
 

Verified Member
Quote:
Originally Posted by echoRausch View Post
Maybe Aleksej should chime in here.

As far as I know RX2 has an algorithm to calculate the "perfect" amount of phase shift to optimize the signal. The result is usually some more headroom.
Yes - in Channel Ops, the preset is called 'Minimize Signal Peak Levels' and it uses the vari-phase option.
I've actually never tried it though.... yet.

from the manual " Vari Phase Rotation - Enabling vari phase rotation will analyze the audio selection and apply the time-variable phase rotation to both left and right channels resulting in a symmetrical waveform, minimizing signal peak levels. "
Old 25th February 2013
  #6
Lives for gear
 
Alexey Lukin's Avatar
 

Verified Member
Quote:
Originally Posted by Nonlinear View Post
If you used a phase rotator on music or voice during mixing or mastering wouldn't that negate the effect, or create some other weird phase effect, if that recording was subsequently broadcast?
Broadcast does not necessarily rotate the phase of music. While this operation does decrease peak levels for highly asymmetric signals (such as speech), it does little to peak levels of music, as music is typically close to symmetric. Anyway, if you apply a 90° phase rotation twice, you'll get a 180° phase inversion — nothing too scary. I think that most broadcast processors do not implement a genuine 90° rotation because it introduces a noticeable latency. They may do some kind of frequency-variable phase shift that usually does a similar job.

The goal of this tool in RX is to provide general phase inversion and phase rotation functionality. Phase inversion can be useful for null-tests, while phase rotation may help you maximize the levels of your music if it is dominated by asymmetric instruments.
Old 25th February 2013
  #7
Lives for gear
 
dcollins's Avatar
 

Verified Member
Quote:
Originally Posted by Alexey Lukin View Post
Broadcast does not necessarily rotate the phase of music. While this operation does decrease peak levels for highly asymmetric signals (such as speech), it does little to peak levels of music, as music is typically close to symmetric. Anyway, if you apply a 90° phase rotation twice, you'll get a 180° phase inversion — nothing too scary. I think that most broadcast processors do not implement a genuine 90° rotation because it introduces a noticeable latency. They may do some kind of frequency-variable phase shift that usually does a similar job.

Here is what broadcast used:

The W3AM Phase Rotating Asymmetry Eliminator!

Allpass paper by W4ENE


DC
Old 25th February 2013
  #8
Gear addict
 

Thread Starter
Quote:
Originally Posted by huejahfink View Post
Yes - in Channel Ops, the preset is called 'Minimize Signal Peak Levels' and it uses the vari-phase option.
I've actually never tried it though....
I tried it on an interview recording and it does in fact "symmetricize" (LOL) the peaks.

And I don't hear any side effects.
Old 26th February 2013
  #9
Lives for gear
 

The phase rotator was originally patented by Lenny Kahn for use in one of his AM radio processors. FCC limitation in AM broadcasting is +125 percent -100 percent modulation. Male announcer speech waveforms can exhibit substantial asymmetry. If you try and heavily clip a waveform like this to control modulation levels, bad things happen. The idea with the phase rotator was shift the phases of the harmonics to reduce the asymmetry to allow clipping to be applied without the adverse side effects. So, it does work well in the intended application. Some users claim that you can hear subtle timbre shifts in the program material with the phase rotation applied.
Old 21st May 2013
  #10
Lives for gear
 
petsematary's Avatar
 

Pardon me for hijacking the thread a bit, but I'm looking for a phase symmetry type tool, too. Am doing mainly spoken word and broadcast stuff and sometime, for various reasons, I'd like to apply a bit of phase rotation, much like the "voice symmetry" option on Symmetrix boxes. Something automated, that will make a track symmetrical in terms of peaks etc. Most phase rotation tools are too advanced for this though. Is there a simpler, itb way of doing this? Thanks.
Old 21st May 2013
  #11
Lives for gear
So it's a good tool to "fix" poorly tracked drums and increase the accuracy of a meter, without having to completely crush something with a super fast compressor or suck the life out of it with a giant notch, pretty much?
Old 21st May 2013
  #12
Lives for gear
 
dcollins's Avatar
 

Verified Member
Quote:
Originally Posted by psykostx View Post
So it's a good tool to "fix" poorly tracked drums and increase the accuracy of a meter, without having to completely crush something with a super fast compressor or suck the life out of it with a giant notch, pretty much?
No, not at all.

Phase rotation was for radio broadcast for reduce the asymmetry in voice.

RX has a phase rotator but I've never used it.


DC
Old 21st May 2013
  #13
Lives for gear
 
huejahfink's Avatar
 

Verified Member
Yes - the vari-phase mode in Izotope RX2 ADV channel ops will do exactly what you ask petsematary
Old 21st May 2013
  #14
Lives for gear
 

I don't care for the effect. It usually involves a Hilbert transformer or a sophisticated dome filter, but for little gain. Usually a proper EQing and lowcutting rids asymmetry.
Old 21st May 2013
  #15
Lives for gear
Quote:
Originally Posted by dcollins View Post
No, not at all.

Phase rotation was for radio broadcast for reduce the asymmetry in voice.

RX has a phase rotator but I've never used it.


DC
Thanks dc.
Old 21st May 2013
  #16
Lives for gear
 
once a roadie's Avatar
 

I've seen quite a few asymmetrical synth and bass waveforms. Anyone care to explain how they become asymmetrical? I've been curious ...
Old 21st May 2013
  #17
Lives for gear
 

It isn't a single phase shift stage - more like 4 series connected allpass phase shifters, each one set to around 360 Hz (for speech).
Old 22nd May 2013
  #18
Lives for gear
 
Alexey Lukin's Avatar
 

Verified Member
Quote:
Originally Posted by once a roadie View Post
I've seen quite a few asymmetrical synth and bass waveforms. Anyone care to explain how they become asymmetrical? I've been curious ...
For many synths this is just a result of certain phase alignment of synthesized harmonics. For natural instruments this often has a connection with a particular way of sound production: when sound is produced in a series of one-sided pulses.
Old 22nd May 2013
  #19
Lives for gear
 
dcollins's Avatar
 

Verified Member
Quote:
Originally Posted by tpad View Post
It isn't a single phase shift stage - more like 4 series connected allpass phase shifters, each one set to around 360 Hz (for speech).
Here are some from a broadcast processor designer:

Free DSP Plug-ins


DC
Old 22nd May 2013
  #20
Gear nut
 
Sempoo's Avatar
 

Waves InPhase, Voxengo PHA-979, iZotope RX2 etc etc
Old 23rd May 2013
  #21
Lives for gear
 
petsematary's Avatar
 

Quote:
Originally Posted by huejahfink View Post
Yes - the vari-phase mode in Izotope RX2 ADV channel ops will do exactly what you ask petsematary
Just dl'd a trial and yes, it does in fact do exactly what I want. Auto-rotation of phase, makes everything perfectly symmetrical. I've heard people say this affects the audio and granted I did not have 100% perfect reference gear when I tried it, but I was unable to hear any difference, it was even as far as the waveform goes but no difference in sound. Which I like, just thought I'd add that if anyone's curious. This was for a voice only track btw. Thanks for the input!
Quote
1
Old 23rd May 2013
  #22
Lives for gear
 
huejahfink's Avatar
 

Verified Member
Quote:
Originally Posted by petsematary View Post
Just dl'd a trial and yes, it does in fact do exactly what I want. Auto-rotation of phase, makes everything perfectly symmetrical. I've heard people say this affects the audio and granted I did not have 100% perfect reference gear when I tried it, but I was unable to hear any difference, it was even as far as the waveform goes but no difference in sound. Which I like, just thought I'd add that if anyone's curious. This was for a voice only track btw. Thanks for the input!
You're welcome mate.
Old 10th September 2013
  #23
Lives for gear
 

Put a square wave into a 90º phase rotator and you get a very spiky signal, with a crest factor of -12dBFS. The square wave isn't asymmetric, but the harmonics are aligned to the space between 1 and -1, and any change of the phases result in a higher crest factor. If we think backwards, this perfect waveform sets the bar for full spectrum dynamic range, because any other signal cannot be as loud (powerful) as it. For all phases, 12dB of headroom will ensure no clipping in one signal, as long as the signal level is gained with congruence to a square wave's harmonic series.
Old 10th September 2013
  #24
Gear nut
 
Conundra's Avatar
 

DMG EQuilibrium has All pass filters that are very useful for this.

They can be very effective for correcting asymmetry in waveforms and increasing headroom if that's what is desired. Particularly good for sawtooth based basslines.

Cheers

Conundra
Old 10th September 2013
  #25
Lives for gear
 
FabienTDR's Avatar
In case someone's interested, a "phase rotator" is in most cases just a very steep phase distortion around the (speaker's) fundamental. That is, a 4-8th order allpass around 200Hz.

There's nothing complicated about it, it's just that this approach works, sometimes (!). But imho, it's no a good idea for mastering. It radically changes the sound of the whole low end and tends to blur the important low mids. Especially the latter makes it pretty useless on full mixes. Phase rotator is a marketing term, better think of it as a phase scrambler. A standard steep Peaking/Shelving filter has a very similar side-effect.
Topic:
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
loglinear / Mastering forum
28
Joemamma / So much gear, so little time!
26
ninjaneer / Music Computers
0
ramjet / So much gear, so little time!
30
loglinear / Music Computers
2

Forum Jump
 
Register FAQ Search Today's Posts Mark Forums Read

SEO by vBSEO ©2011, Crawlability, Inc.