Can someone explain this? Asymetrical waveforms . . .

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I hope this article isn't going too far off-topic.
I think it's interesting to see all the types of signal
manipulation that go into broadcasting audio signals.
There is a section on the use of the "phase rotator"
(a chain of all-pass filters)

http://tinyurl.com/hv2ef

Cheers.

ZAP
 
bcarso said:
Some good examples of very unsymmetrical waveforms can be found on some of the minimally-processed Sheffield recordings, ideally one of the direct-to-disc LPs of Harry James.  However the asymmetry of James' trumpet and one of the trombone solos survives nicely on the CD.

These are, as noted, hell on wheels for potentially clipping things throughout the signal chain and of course present terrible problems for broadcasting if not alleviated.  But oh my goodness do they add to the realism if accurately reproduced!

These sorts of material are likely to allow absolute polarity detection, although what that perception is due to is questionable---is it the effect the signal is having on the system, on your own ears/brain (the so-called Wood* effect), or both?

A little story I may have related elsewhere:

When the UCLA Electronic Music Studio director decided he had to have Dolby B noise reduction on a tight budget (this was circa 1969), he purchased some consumer boxes (branded Advent IIRC?).  They worked o.k. for some program material and horribly for others.  It turned out that the polarity of specific waveforms common to the synthesizers made an enormous difference.  Internal to the boxes were amplitude detection circuits that, to save a few nickels were strictly half-wave.  They and the level control circuitry they served were faked out almost completely if the sawtooth had a slow rise and rapid fall versus the opposite.  You get what you pay for.  And for most heavily processed consumer audio the boxes worked fine.  For amateur recordists I wonder though.

To save the day, in collaboration with a physics grad student who was an amateur musician and hung around the studio, we devised an add-on that would not per se require surgery on the boards, but would merely be tacked on in parallel to make the detector circuits symmetrical.  It worked.  The downside was that the edge connectors for the stock boards were unbelieveably cheap and good for at most a few mating cycles, so the result was a plague of intermittencies.

As so often was the case, the engineers were initially praised and ultimately reviled.  After dealing with the prima donnas and getting completely fed up, I quit to work full-time for the Astronomy Dept., where I remained until 1985.


*no relation that I know of

I actually have a copy of "the wood effect" book (by R.C. Johnsen) but to be honest I never forced my self to read beyond page 22, where I am mentioned favorably for stating that absolute polarity was audible in print back in 1983.  From skimming the book, the author dimes on several well know people who were dismissive of absolute polarity. Some very recognizable names that I suspect might be embarrassed today.

A Mr. Charles Wood is credited with discovering the audibility of absolute polarity back in 1957. I'm not sure it has ever been widely embraced and properly understood by lay people.

JR
 
Hi Zapnspark,


  thanks for the last link too! very interesting indeed. Very well worth reading. I just learned a lot. Many thanks! Now I just have to persuade my clients that louder isnt necessarily better, which should be much easier armed with this document . . . .

quote - "Phase rotator. The phase rotator is a chain of allpass filters (typically four poles, all
at 200Hz) whose group delay is very non-constant as a function of frequency. Many
voice waveforms (particularly male voices) exhibit as much as 6dB asymmetry. The
phase rotator makes voice waveforms more symmetrical and can sometimes reduce the
peak-to-average ratio of voice by 3-4dB. Because this processing is linear (it adds no
new frequencies to the spectrum, so it doesn’t sound raspy or fuzzy) it’s the closest
thing to a “free lunch” that one gets in the world of transmission processing. "


. . . . so, if the phase rotor reduces the peak to average ratio of voice by 3-4dB, if I construct a filter of me own, would it help my mixes when they get to the radio? would processing of the lead vocal under my control in the studio be preferential to it being done do the whole mix remotely in a radio transmitter station? . . . .3-4dB is quite a lot methinks . . .


  Kindest regards,


    ANdyP
 
Mbira said:
I know that "phase correction" is one of the things that mastering engineers often spend their time doing.  Does this often involve trying to "correct" some of these naturally occurring anomalies? 

For instance, that Harry James reference-would a modern Mastering engineer try and "fix" that?

I doubt that they would unless they were concerned about overload.  If one were recording sufficiently below 0dB (or below an acceptable amount of tape saturation with an analog tape recording) it wouldn't be a huge issue.  But for radio broadcasting it might well be addressed.

I must say I was shocked when I saw the waveforms on an oscilloscope.  I thought the equipment under test was malfunctioning, until I verified that it was right there in the source.

One of the Sheffield's liner notes actually talks about how they had to reconvene the whole session after the mastering engineer realized that things had not gone well enough with the lathe.
 
strangeandbouncy said:
would processing of the lead vocal under my control in the studio be preferential to it being done do the whole mix remotely in a radio transmitter station? . . . .3-4dB is quite a lot methinks . . .


from the orban article:
Manufacturers have tuned broadcast processors to process the clean, dynamic
program material that the recording industry has typically released throughout its
history.

It's very sad that program material these days tends to get tuned to broadcast processors
 
strangeandbouncy said:
What is on my mind is doing 2 "masters" of each track. One for CD, and 1 for Radio . . . .

I don't think that would be a good idea. You would be doing the process twice.
Like mentioned in the article, heavily compressed masters won't sound better than dynamic masters after orban processing.
Your song will sound "heavily compressed, times two"
I wonder what "4pole bandpass filter around 200hz times two" sounds like
 
Arno,


  don't think you quite follow me. I am talking about preparing a separate, more open, less limited version for radio only. Absolutely twice, but only mastered once each! My clients all love huge, fat mixes, so I do as much in the analogue domain as possible, including a fabulous NTP brickwall limiter, adding as little digital limiting as possible to reference mixes, and submitting mixes without "finalising" limiting to master. Whilst it may help initial approval of my mixes to flatten as much as possible, it might not help my success with radio mixes from what I have read here. I wonder if using a phase rotor on the vocal(especially if as wonky as the one I posted earlier), earlier in the process would help matters further down the change or not. It is rather a risky science for me to try, and wait 'til I hear my work mangled on the airwaves . . .

If using a phase rotor changed the sound of the vocal, I would rather have control of it at my end . . .

  Kindest regards,


  ANdyP
 
Coming at it from the other direction...there is no reason why asymmetric waveforms should not be possible in nature.  I just built a simple spreadsheet to allow me to sum a few sinusoids with given freq, phase, and magnitude.  Just fiddling with four frequencies I was able to produce two highly asymmetrical waveforms.  Pics below.  The spreadsheet also corrects for DC offset, so the graph y-axis is correct.

Asymmetry 1

Asymmetry 2

If someone is curious I can upload this new spreadsheet which is a little different than the other one I posted a couple of years back when we had a related discussion in The Drawing Board about harmonic content.

Are these pics showing up?  I'm not seeing them here...

A P
 
Not showing up here either AP.  The dreaded red x...

One thing on this subject that just occurs to me (and perhaps has been considered before):  how feasible would it be to do, in the phase domain, a dolbyesque sort-of complementary processing?  That is, you do the shifting about preserving amplitude response but reducing the peak-to-average shape of the instantaneous waveforms, so as to allow optimal radio transmission.  Then you undo it at the receiving end, using some sort of maximum-entropy-like algorithm, some inference engine detecting the likely instrumentation, or even some subcarrier information guidance.

In Dolby the system works (when it does work) by knowing that low-level signals have been boosted a given amount, so that the process can be un-done deterministically at playback to recover the dynamic range.  Of course the pitfalls are also well-known, as the amplitude detection process can only respond so quickly, so the complementation is only partially successful in preserving attack transients.

With Dolby of course unless you know the material is so encoded you get frequency response anomalies by not enacting the complementary processing on playback.

Just a thought.
 
Blimey!


  My simple question about some wonky vocals has led straight into my biggest bugbear(radio mastering) without my having made a connection . . . .


  . . . . . and now some serious grey matter stimulation . . . bcarso, if I follow your drift, you'd better patent or copyright that idea before someone else does . . . something that fits BOTH criteria, max broadcast levels/AND max fidelity . . .



    fascinating stuff!


    Thank you!


    ANdyP
 
strangeandbouncy said:
Blimey!


   My simple question about some wonky vocals has led straight into my biggest bugbear(radio mastering) without my having made a connection . . . .


   . . . . . and now some serious grey matter stimulation . . . bcarso, if I follow your drift, you'd better patent or copyright that idea before someone else does . . . something that fits BOTH criteria, max broadcast levels/AND max fidelity . . .



     fascinating stuff!


    Thank you!


    ANdyP

Too late now Andy, but thanks for the appreciation!

As long as a community of minds can be stimulated, Bose can't patent it, and I have enough to eat from the occasional bit of work, I'm satisfied within reason.
 
AP,


  that's some more fascinating info. Thank you very much!



  much re-evaluation of my work practices going on here, and I an rying to organise a beer with a BBC transmission engineer who is a friend of a friend to discuss further.


  next question is, what do I need to know about DIGITAL broadcast processing., which will be here very very soon. Will that open another can of worms as far as mastering is concerned?


    Kindest regards,



      ANdyP
 
I am talking about preparing a separate, more open, less limited version for radio only. Absolutely twice, but only mastered once each! My clients all love huge, fat mixes, so I do as much in the analogue domain as possible, including a fabulous NTP brickwall limiter, adding as little digital limiting as possible to reference mixes, and submitting mixes without "finalising" limiting to master. Whilst it may help initial approval of my mixes to flatten as much as possible, it might not help my success with radio mixes from what I have read here.

At least here radio versions are done for important singles only and by and for the guys who know. There seems not to be resources to process and manage whole albums for radio only. Not that it is a lot of work though. My friends at Chartmakers Mastering routinely leave 2-3 dB more dynamics for radio. Less compression and no brick wall limiting. Quite soon people noticed that the songs come out of the radio as loud as everything else (I guess Orban processors are in use here) and _sound better_ Of course the radio processor is not as cool as some high end clipper in mastering, but it is just that the order of processing works much better.

-Jonte

 
Another thing to be aware of when viewing digital audio data is that some tools are better than others.  I used to use Sonar and at the time (v3.x and 4.x) its waveform envelope display was not accurate.  Its seems that one of the SW guys there just took the pos (or neg) half of the data and replicated it with a neg sign to more quickly generate the waveform envelope displayed in each track.  They were always symmetrical until you zoomed in far enough that it started to show individual samples at which point it was accurate, but by then your viewable context of perhaps around 1000 samples, about 1 cycle of 40Hz at 44.1kHz, was small enough that you didn't really notice the asymmetry. 

I've since switched to Reaper (for other reasons) and it accurately displays the asymmetry in the envelope regardless of zoom.  This same topic came up in the Reaper forum sometime last year or so when someone noticed asymmetry in something they had recorded.  I've forgotten whether it was vocals or a wind instrument recording.  Perhaps this problem has been fixed in other tools, but I wouldn't count on it.

A P

EDIT:  I mis-remembered--it was SAW Studio that displayed inaccurate waves when zoomed out.  Below is a link with comparison displays:

http://forum.cockos.com/showpost.php?p=128232&postcount=12
 
AnalogPackrat said:
Another thing to be aware of when viewing digital audio data is that some tools are better than others.  I used to use Sonar and at the time (v3.x and 4.x) its waveform envelope display was not accurate.  Its seems that one of the SW guys there just took the pos (or neg) half of the data and replicated it with a neg sign to more quickly generate the waveform envelope displayed in each track. 
A P

If true that is almost all-surpassingly stupid.  Reminds of of some advice a group I was working with got from the then-chief administrator for NSF funding for astronomy, about truncating data rather then doing an accurate roundoff.  Apparently the systems he was accustomed to working with were noisy enough that this made no difference.  Idiocy.
 
Sad, but true.  However it was SAW Studio, not Sonar.  My mistake.  I've edited my previous post and put a link there in case you are interested. 

My profession is computer vision--unfortunately I have to do a lot of my work in software even though "solder is my favorite programming language."*  I've spent a lot of my career in the semiconductor and flat panel display capital equipment industries.  You'd be surprised at some of the absolute crap that goes out the door in equipment costing $1-5M per unit.  I don't know what's worse, people who should know better or people who just don't have a clue and yet are put in a position (usually by some fancy pants manager) where they can sow their foul seeds of stupidity throughout the system.

A P

* I believe that is attributable to RAP.
 

Latest posts

Back
Top