The AM Forum

THE AM BULLETIN BOARD => Technical Forum => Topic started by: N1BCG on May 16, 2023, 11:26:59 AM

Title: A Closer Look At Your Voice
Post by: N1BCG on May 16, 2023, 11:26:59 AM
It doesn't seem to make sense, but it does...

I'm researching voice characteristics and would like to get some input from willing ops who can monitor the asymmetry of their voices. Specifically, I'm trying to determine the percentages of those who have a consistent polarity (asymmetry is always in the same polarity regardless of speech) or have mixed polarity (where the asymmetry is different for lower vs higher vocal sounds).

In other words, are there moments where either the negative or positive modulation level drops even though the transmitter seems fully modulated? Lower vocal sounds "Aaaahhhh" and higher vocal sounds "Eeeeeee" can be 180 degrees apart, resulting in microphone polarity to be correct "mostly" but not "always". In this case, a low frequency equalization cut or boost can flip the dominant polarity of one's voice. It's quite fascinating.

A modulation monitor that can simultaneously display both negative and positive modulation would be ideal, but a look at either will reveal a drop when the asymmetry favors the opposite polarity.

I've seen all of this from a receiver, but audio processing and propagation will skew the results, so obtaining the characteristics at the transmitting end with processors set to "proof" (or bypassed) will be more accurate. Your input on this would be greatly appreciated!

Title: Re: A Closer Look At Your Voice
Post by: K1JJ on May 17, 2023, 10:39:57 AM

For now, we won't debate the pros and cons of all-pass phase rotators and symmetrical audio -  and rather talk about big positive peaks and asymmetrical audio above 125%+ positive using a good performing AM diode or sync detector.

I see pretty much the same as you when it comes to audio polarity effects.  We can definitely change the polarity of our voice when it is run thru an EQ or thru a transmitter that is not flat. Polarity is most sensitive in the extreme lows.

The classic scope "shark fins" polarity look in the lows requires a very clean and flat transmitter system especially in the extreme low end.  I have six homebrew plate modulated transmitters (all have mod transformers with negative feedback) and none of them show perfect low end shark fins because they are all flat (and clean) only down to about 20 Hz. The exception is the series modulated 813 X 813s which is flat down to 1 HZ. This rig looks like a class E or SDR rig.  Even though the voice has little energy below 30 Hz, it still shows up as shark fins once we boost the EQ extreme lows.    I'm having a ball running 180%+ audio with the class E guys when it's running.

For example, like most of us, Steve/QIX's voice does not have much low end below 60 Hz, but after his processing, it contains a very nice balance of extreme lows. The polarity is great, like over 150% positive and more.  Bob/ KBW sounds FB as well. If either used a plate modulated rig with a mod transformer that rolled off the lows below 30 Hz, the shark fins would disappear as well as the enhanced sound.   The REA mod monitor is a great way to monitor both sides.

As for the highs above 200 Hz, 3 KHz, etc, the EQ will also play a big factor in polarity optimization.  My series modulated 813 rig performs best of all the rigs here too. Again, a flat and clean transmitter not colored by imperfections is needed to be able to manipulate the best polarity.

We once ran an audio polarity test with a group of AMers in person.  This was pure audio from a mic into a scope.  Of five guys, the polarity ranged from about 110% positive up to about 140% positive.  The 140% guy had a classic powerful AM broadcash voice. The 110% guy had the usual mid-rangey voice that needed EQ to sound good on the air. Yes, we could actually hear the sound of asymmetry in the voices, in person.

I think there IS a place for audio polarity experimentation. It is a personal thing and some voices can benefit from it and some cannot.  A first step is to find out what your own voice polarity looks like with a simple mic and scope and go from there.   A big peak at 80 Hz may produce a BIG Barry White sound; while a peak at 200 Hz may give a poor, mid-rangey, unwanted muffle. Optimized audio enhancement to fit our particular voice is an art form.


Title: Re: A Closer Look At Your Voice
Post by: N1BCG on May 17, 2023, 11:29:49 AM
Thanks for the response, Tom. The purpose of my asking is strictly to draw attention to the complexity of our voices when determining "polarity" from mic to modulator, and that's mostly important to those who want to maximize their asymmetry.

A key observation I've made is that the dominant polarity is *not* uniform throughout the vocal spectrum for many of our voices. Positive peaks will exceed negative peaks most, but not all, times. When this happens, the dominant negative peaks will either force excessive gain reduction in the full-wave detectors of a compressor and/or the negative clipper will get slammed with energy, potentially causing spatter while the positive peak meter won't reach 100% at those moments.

Everything can look good on a modulation monitor, mostly, and yet there can still be splatter caused by these polarity reversals.

Complicating the matter, equalization will also change the dominant polarity of a voice. As an example, boosting low frequencies will flip the dominant polarity in that range which can cost the op in loudness.

I've been experimenting with a solution that uses, incredibly, an all-pass filter (phase rotator). However, instead of an eight pole design to create a symmetric waveform, a two-pole filter is used to invert the lower vocal range to match the higher vocal range.

That, and employing a half-wave detector for compression which ignores the positive half of the audio waveform. Unlike the fulll-wave detectors used in nearly every off the shelf compressor (including many favorites used in hamateur radio), a half-wave detector will maintain a maximum negative % modulation regardless of vocal asymmetry.

Employing both creates an ultra-clean signal with relentless positive peak energy.

I think the topic is worth discussion.

Title: Re: A Closer Look At Your Voice
Post by: K1JJ on May 17, 2023, 11:48:15 AM

Very interesting about designing a "dynamic" device to better manage the polarity swings.   And, if it wasn't enough to manage already, there are days when my voice changes to the point where the previous EQ/ polarity settings are all wrong.  My positive peaks go to hell until later in the day when the vocal chords warm up.... :-)   Which also means my settings will probably not work for you, etc.   Like trying to swap shoes.

Bottom line is dominant polarity isn't constant thru the full spectrum and changes around to boot.  It's like having a bad hair day, but knowing it will all come back to the norm (requiring the original enhanced settings) later on.

Keep us informed with the experiments.


Title: Re: A Closer Look At Your Voice
Post by: DMOD on May 18, 2023, 11:19:24 AM
I have measured the average asymmetry of my voice and it is 65% positive and 35% negative.

Steve's WA1QIX's AM monitor will show any asymmetry in the RF envelope as well.

Phil - AC0OB

Title: Re: A Closer Look At Your Voice
Post by: N1BCG on May 18, 2023, 01:28:49 PM
That generally seems to be the norm and it's likely that either cutting or boosting frequencies below 180Hz will significantly change that ratio.

I've found that upper-mid frequencies just below the "presence range" tend to define predominant voice polarity although aggressive low and high equalization can cause inversions. It's not uncommon to have sibilant highs ("Ssssssss") invert and slam a negative peak limiter (a.k.a. clipper) with the resulting square waves causing splatter even without hitting the baseline.

Using a two-pole inverter to flip lower frequencies and a simple pre-emphasis circuit (not 31 band equalization) virtually eliminates negatively polarized asymmetry. At least that's what the tests are showing.
AMfone - Dedicated to Amplitude Modulation on the Amateur Radio Bands