What kind of terrible watermarking is this (KFI and others)?

[yeoldeschool]

1 minute 20 second excerpt from Coast 2 Coast AM's 2 AM PST hour (11/14/2022) on KFI as captured from an SDR receiver:

https://files.catbox.moe/r53ema.wav

Spectralgram:

https://i.postimg.cc/W4pMcWHz/KFI-voltair-insane-11-14-2022-2-AMhr.png

Can anybody identify this terrible sounding watermarking technology?

I'm under the impression that Voltair is the only ratings watermark enhancement processor in common use. That being said, at 25-Seven Voltair Watermark Monitor & Processor | Telos Alliance, its makers ask "Can competitors or Nielsen tell if I’m using Voltair?", followed by answering themselves with, "By design, Voltair leaves no fingerprints, as most audio processors don’t. So we don’t think so." This leads me to believe that it's impossible for anyone to actually hear it. And yet there it is, sounding vaguely reminiscent of the "ice cream truck" filter present in CoolEdit's old Notch Filter transform.

Is this stuff truly Voltair? Or is it something else? Is it being modified against the technology creators' wishes to make it this audibly obvious, or is this really how it's designed to sound? I cannot understand how even the most deaf of listeners aren't being driven away in numbers greater than whatever gains they're making by inserting the watermarking loudly enough to trigger every last PPM device possible. It sounds like the entire station is being run through a miniature physical echo chamber with dampered strips of zangy-sounding sheet metal and springs vibrating to the music and speech. You can even see the watermarking in this audio clip's spectralgraph, looking like Freddie Kruger's personal claw marks (for all 10 fingers no less) scratching their way along in the time domain over its entire length. When I first noticed this sound appearing on numerous stations, I actually thought my tuner's caps and other components were croaking, and that they were producing some bizarre form of ringing.

Incredible.

 

[SomeRadioGuy]

Thats voltair and i havent even listened.. its a metallic ringing turd yoiu cant flush.. its supposed to watermark the audio and get the most out of it for PPM ratings.

 

[DavidEduardo]

I'm under the impression that Voltair is the only ratings watermark enhancement processor in common use.

Nielsen has its own enhancement box.

That being said, at 25-Seven Voltair Watermark Monitor & Processor | Telos Alliance, its makers ask "Can competitors or Nielsen tell if I’m using Voltair?", followed by answering themselves with, "By design, Voltair leaves no fingerprints, as most audio processors don’t. So we don’t think so." This leads me to believe that it's impossible for anyone to actually hear it. And yet there it is, sounding vaguely reminiscent of the "ice cream truck" filter present in CoolEdit's old Notch Filter transform.

When turned up too high, Voltair interprets background noise, pauses in songs that have some kind of noise, etc., to be places to insert the coding. That creates strange artifacts because it is detecting insertion points that are too "low".

Is this stuff truly Voltair? Or is it something else? Is it being modified against the technology creators' wishes to make it this audibly obvious, or is this really how it's designed to sound?

No, it is individual stations setting it annoyingly high. Because talk formats have vastly fewer places to insert codes, they seem to overdo it a lot.

I cannot understand how even the most deaf of listeners aren't being driven away in numbers greater than whatever gains they're making by inserting the watermarking loudly enough to trigger every last PPM device possible. It sounds like the entire station is being run through a miniature physical echo chamber with dampered strips of zangy-sounding sheet metal and springs vibrating to the music and speech.

That sounds, in part, like the narrow bandwidth setting of a pure digital SDR contributing to the effect. Many of them have very active AGC circuits as they are designed to lock and hold distant signals with variable strength.

You can even see the watermarking in this audio clip's spectralgraph, looking like Freddie Kruger's personal claw marks (for all 10 fingers no less) scratching their way along in the time domain over its entire length. When I first noticed this sound appearing on numerous stations, I actually thought my tuner's caps and other components were croaking, and that they were producing some bizarre form of ringing.

Do you hear this on a "normal" radio, or just your SDR or a distant SDR you hear in a stream?

 

[DavidEduardo]

Thats voltair and i havent even listened.. its a metallic ringing turd yoiu cant flush.. its supposed to watermark the audio and get the most out of it for PPM ratings.

All depends on the setting of the Voltair. Too many stations set it way too high and end up making the audio very fatigue-laden and hard to hear for more than a short time.

 

[PTBoardOp94]

That blurb from Telos about the Voltair being transparent is hilarious.
If no one, not even Nielsen, can tell you're using Voltair, then it can't possibly work. The voltair leaves artifacts as a necessity of its function.

It's like lighting a cigarette in a bar and assuming no one will notice.

EDIT: To OP's question ... is that a Voltair? I don't know for sure. I've never heard encoding that intense before.

 

[DavidEduardo]

That blurb from Telos about the Voltair being transparent is hilarious.
If no one, not even Nielsen, can tell you're using Voltair, then it can't possibly work. The voltair leaves artifacts as a necessity of its function.

That is not true. The Voltair simply enhances at a level each user determines the tiny frequency ranges where encoding can happen. The encoding is, then, masked by the higher level audio and does not distort it because the code is a digital pulse.

Audio is made fatiguing when the Voltair is set to enhance those coding frequency bands too much, making things like background noise and low level content in a music selection be overly amplified to mask the PPM code burst.

It's like lighting a cigarette in a bar and assuming no one will notice.

No, it's not. The overly-active Voltair enhances station audio so that it will mask more opportunities for code insertion. The code remains the same. You can't hear the code. What become noxious is station audio that is processed too much to open coding opportunities.

 

[PTBoardOp94]

That is not true.

No. The things you describe are things that can be measured, which contradicts the statement from Telos that "we don't think Nielsen can tell you're using Voltair". Nielsen, or anyone else so inclined, can make measurements and figure it out.

Measured is different than "can be heard with the unaided ear." I probably should have used more precise verbiage.

 

[DavidEduardo]

No. The things you describe are things that can be measured, which contradicts the statement from Telos that "we don't think Nielsen can tell you're using Voltair". Nielsen, or anyone else so inclined, can make measurements and figure it out.

No, what the Voltair is doing is changing the composition of a station's program audio to make the PPM injection frequencies "louder" so there will be more opportunities to inject the PPM code in proportion to the station's program audio. To detect Voltair, you'd have to compare the actual program content with the unprocessed original audio. In the case of talk programming, that is impossible or improbable.

 

[yeoldeschool]

I think I understand what David is saying. The Voltair acts like a super-fine 100 or maybe even 200 band EQ, where only the 10 specific hair-thin bands into which the PPM encoders will place their data get amplified. If amplified a lot, those tiny highly vertically-skirted bands being amplified excessively versus their surrounding audio creates audible filter ringing. Except the ringing isn't continuous, since the amplification only happens upon the presence of sounds. So you hear ringing when there are sounds, but no ringing when there aren't. (I have always noticed I could create ringing in CoolEdit by using the FFT and the correct precision filters -- like Blackmann-Harris at 24000 bands. If I created extremely thin bands -- imagine one slim enough and vertically skirted enough to notch only a CRT whine out of audio, but locating that band at 1 kHz instead, and amplifying excessively instead of attenuating ... well, something that narrow rings if you make too great a volume change). Do I have it, David?

As far as KFI:

That sounds, in part, like the narrow bandwidth setting of a pure digital SDR contributing to the effect. Many of them have very active AGC circuits as they are designed to lock and hold distant signals with variable strength.

Do you hear this on a "normal" radio, or just your SDR or a distant SDR you hear in a stream?

I hear it through my own analog tuners, live off the L.A. airwaves, with exactly the same intensity audible in that clip. It truly is so bad, the first time I heard it happening at this level, I assumed the components in my radio were dying.

(edit: And the tuners I heard it through at that level, and can still hear it through at that level, are all 80s and 70s gear, so there's no way any fast dynamics processing is present in them, either.)

About the SDR, the one I used does not do any fast compression, only RF AGC, which you can actually disable in the UI. You can find a list of hosts running the software at http://kiwisdr.com/public/ (the one used for the particular WAV above was http://kphsdr.com:8072/ at 2 AM when the DX reception was excellent).

 

[sdwulfdawg]

Can one actually hear the so-called "watermarking"? This conversation thread has me really intrigued.
What does it sound like?

 

[yeoldeschool]

David says you can't hear the PPM watermarking in music/speech -- that you can only hear the effects of the Voltair giving the watermarker more slices of time in which to place the watermark data (those additional slices of time producing audible effects if driven too hard).

The only way to hear a watermarker would be to invert and cancel the pre-marked audio against the watermarker's output. That can't be done without a willing engineer at a station with a PPM encoder who could capture bit accurate digital audio from the "before" and "after" points surrounding the watermarking box. (I'd be curious to hear what it sounds like too, if any such engineer is out there reading.)

 

[SomeRadioGuy]

Can one actually hear the so-called "watermarking"? This conversation thread has me really intrigued.
What does it sound like?

i joked, but its a metallic rinnging digitally sound....

 

[SomeRadioGuy]

All depends on the setting of the Voltair. Too many stations set it way too high and end up making the audio very fatigue-laden and hard to hear for more than a short time.

then every station that has it must have it set in correctly.. i heard it on WMJI 105.7 cleveland when i lived in NW PA and it was AWFUL... downright disgusting and obvious to even the average listener

 

[Theater of My Mind]

I remember when the actual on-air product used to be the priority at radio.

 

[TheBigA]

I remember when the actual on-air product used to be the priority at radio.

Shouldn't every listener count? Isn't that important too? You might have the best programming in town, but if Nielsen isn't counting your listeners properly, that will mean less money for programming.

Running a radio station isn't as easy as it once was. They once said it was a license to print money. Not so much now. It's even harder when the company that's responsible for the ratings makes it harder to get an accurate count.

 

[Yabadabado1]

I am intrigued. I got out of radio before PPM was a "thing". So, what's the purpose of making it "louder" so to say? To make sure PPM meters are picking it up from longer distances? Like, if it was set "normally"
the PPM meters within so many feet would register it, but making it louder means the PPM meters could pick it up a couple of thousand feet away?

 

[DavidEduardo]

then every station that has it must have it set in correctly.. i heard it on WMJI 105.7 cleveland when i lived in NW PA and it was AWFUL... downright disgusting and obvious to even the average listener

Not every station in PPM markets uses a Voltair. Some use the Nielsen device that won't allow extreme artifact creation, and some don't use any enhancement. Among competitive stations, there is the fear that if you don't "turn it up to 11" your competitor has an advantage. On the other hand, stations wonder why TSL is lower than ever and I personally think that the Voltair is the reason.

 

[DavidEduardo]

I am intrigued. I got out of radio before PPM was a "thing". So, what's the purpose of making it "louder" so to say? To make sure PPM meters are picking it up from longer distances? Like, if it was set "normally"
the PPM meters within so many feet would register it, but making it louder means the PPM meters could pick it up a couple of thousand feet away?

It's not about "loudness". The PPM encoder at each station sends a roughly 4 second long data burst, up to about 12 times a minute. But to do that, there has to be adequate audio on one of the group of frequency mini-bands where the encoder can generate the data burst in order to mask that data.

If there is not audio of sufficient audio to encode, the Altair looks for the right frequencies but at too low a level, and then boosts one of those tiny pieces of the spectrum to enable the encoder to think there is sufficient masking audio.

If the Altair is set to high, it will boost audio that sounds awful... background noise in a studio like the air conditioning or vibrations in the recording studio or instruments that were not meant to be mixed "hot". The end result is the introduction of unnatural sound.

The overall audio of the station is not louder than normal... just tiny little pieces are raised loud enough for the PPM encoder to activate, resulting in more data bursts per minute being possible.

 

[Yabadabado1]

It's not about "loudness". The PPM encoder at each station sends a roughly 4 second long data burst, up to about 12 times a minute. But to do that, there has to be adequate audio on one of the group of frequency mini-bands where the encoder can generate the data burst in order to mask that data.

If there is not audio of sufficient audio to encode, the Altair looks for the right frequencies but at too low a level, and then boosts one of those tiny pieces of the spectrum to enable the encoder to think there is sufficient masking audio.

If the Altair is set to high, it will boost audio that sounds awful... background noise in a studio like the air conditioning or vibrations in the recording studio or instruments that were not meant to be mixed "hot". The end result is the introduction of unnatural sound.

The overall audio of the station is not louder than normal... just tiny little pieces are raised loud enough for the PPM encoder to activate, resulting in more data bursts per minute being possible.

OK, now I understand [sorta].

 

[DavidEduardo]

Long and detailed, here is a thorough description of PPM encoding, the effects of "enhancing" the encoding and the use of PPM in netcasting:

https://www.indexcom.com/tech/ppm/

"Recently, devices that increase the PPM signal level to audible levels have become available. Their goal is to capture more PPM receivers, increasing ratings by essentially gaming the system. Making the tones audible breaks Nielsen Audio’s carefully designed psychoacoustic masking model. Fundamentally, this is very different from increasing the amount of audio processing to achieve a higher probability of masked PPM encoding, because these devices smear transients and unmask annoying, program-modulated inharmonic tones in the frequency range to which the ear is most sensitive.

This is a dangerous and slippery slope. Put simply, these devices make your radio station sound worse, trashing broadcast audio, driving more listeners away from traditional broadcast in favor of new media and streaming that isn’t using such devices. With this philosophy, you must hope to gain more audience than you will lose. Because new media and streaming does not suffer from pre-emphasis-induced high frequency headroom limitations and is impervious to noise and multipath distortion, new media and streaming already has a sonic advantage over FM radio. Why handicap FM further? Proceed with extreme caution here!"