Audible PPM Signals

[k6sti]

For quite some time I've been hearing what I think are PPM encoding signals on KKJZ 88.1 and KCRW 89.9. They were louder on KCRW until a couple weeks ago, when the level dropped. They are still audible but not as obtrusive. The signals are very audible on KKJZ, particularly on the announcer's voice. They were quite loud on KPCC 88.3 for over a year until I complained to Lance Harper and he did something to tame them. I hardly ever listen above 92 MHz so I don't know if I can hear the signals on other L.A. stations. Anyone else hearing this stuff?

Brian

http://ham-radio.com/k6sti

 

[Hot Hits]

Have not heard them, don't listen to Terrestrial Radio anymore, but that does remind me of the Secondary & Tertiary Cue tones that could be heard with the old style Cart machines

 

[Kelly A]

They may be running one of those Voltaire boxes from Telos Omnia to jack up their Nielsen encode data. If they aren't set correctly you can hear the data. Voltaire, another way for stations to race to the bottom.

 

[k6sti]

A cranked-up Voltair is what I suspect, Kelly. But I don't understand why a public station would sacrifice audio quality in an attempt to boost ratings. I suppose they need ratings to attract underwriters, but I don't believe that is a major source of funding for most public stations.

I heard the funny sound on KPCC for over two years. That may have included time prior to the availability of Voltair. I kept e-mailing Lance Harper about the issue, but the sound persisted. Finally, I e-mailed a recording of some particularly egregious instances and the problem went away the next day. Perhaps I should e-mail a recording to KKJZ. The PPM signal, if that's what I'm hearing, is so strong that it is even audible on breath noise and background studio noise. It's much less obtrusive during music, but once your ear becomes attuned to the characteristic PPM twanginess, you can hear it on just about any source material. I have heard music on KKJZ fade away into a background of pure, continuous twang.

Here are two brief samples of the KPCC sound I recorded in June:

https://www.dropbox.com/s/7g5wd5sbeae46ae/ppm1.wav?dl=1
https://www.dropbox.com/s/om30cc3ngxn0yjx/ppm2.wav?dl=1

Brian

 

[davideduardo]

A cranked-up Voltair is what I suspect, Kelly. But I don't understand why a public station would sacrifice audio quality in an attempt to boost ratings. I suppose they need ratings to attract underwriters, but I don't believe that is a major source of funding for most public stations.

I heard the funny sound on KPCC for over two years. That may have included time prior to the availability of Voltair. I kept e-mailing Lance Harper about the issue, but the sound persisted. Finally, I e-mailed a recording of some particularly egregious instances and the problem went away the next day. Perhaps I should e-mail a recording to KKJZ. The PPM signal, if that's what I'm hearing, is so strong that it is even audible on breath noise and background studio noise. It's much less obtrusive during music, but once your ear becomes attuned to the characteristic PPM twanginess, you can hear it on just about any source material. I have heard music on KKJZ fade away into a background of pure, continuous twang.

If the sound is identical each time you hear it, across different voices and audio content, it's likely the problem is not the PPM encoding. PPM tags can be broadcast on one of multiple frequencies in a fairly broad range of the lower audio spectrum which is several kHz wide. The encoder picks the best frequency based on the audio content and encodes at the frequency that provides the best masking.

The encoder requires a base level of audio to encode... enough that the audio will fully mask the PPM tag. That's why multiple frequency options are broadcast. The tag itself is a data burst which can be created at any particular one of the different PPM frequencies.

Only a Voltair cranked way up might encode when it should not. But I don't believe even a full-blast Voltair can do what you describe. The usual noxious effect is distorting the audio as the Voltair selectively increases particular audio frequencies so as to insert a tag at a higher level. This can have the "badly adjusted EQ" effect on the audio. Your audio samples seem to occur only at one frequency... but that could be notable in those segments because of the narrower range of a particular voice.

If you get the effect on different audio (mail, female voices, music or other non-vocal sounds) it might be the new CBET encoding which was rolled out a few weeks ago in LA and, around the same time in other markets. Noxious artifacts have been allegedly noted in several markets, including Denver and Seattle. Some theorize that this is the effect of the CBET encoding in areas prone to multipath. [/QUOTE]

 

[Kelly A]

After listening to the audio files you posted, I agree with David. It doesn't sound like Nielsen encoded data, but could be one of several things: If they're using audio via IP audio consoles and you over-drive them, the distortion sounds different than analog or AES audio before it breaks up completely. Assuming this recording is from an analog receiver and not their HD stream, it could be something wrong with their audio processing (including a poorly set up Voltaire) or involving the audio program delay used to delay the analog audio to match their HD audio. Do you hear this sort of thing with music too?

 

[k6sti]

The audio samples I posted are from KPCC in June 2015. They are the ones I sent to Lance Harper. Here is his response:

"We're on a new transmitter system as of yesterday and I'm sure I know where in the air chain where the problem is at. It should sound good tomorrow.

The PPM encoder is introducing distortion :-( I'll swap it out tonight."

KPCC's signal did clear up the next day. However, since I had heard the sound for at least two years, I didn't see how it could be due to a "new transmitter system." I also found an e-mail that I sent to Lance in March about the sound appearing on KUOR 89.1, KPCC's outlet in the Inland Empire. It was there for a day and then disappeared. I describe it as the same sound I heard on KPCC. I have not heard it since on KUOR. I rarely can hear KPCC these days so I don't know how it currently sounds.

The sound varies greatly with program material. It is far less audible on music than voice. Although its level differs, the sound character is very similar among the stations I've heard it on. Yesterday I tuned around a bit above 92 MHz. I think I'm hearing the sound on KIIS 102.7, KOST 103.5, and KBIG 104.3. I need to listen more to be sure. It is not as obtrusive on these stations as on KKJZ or KCRW, or as it used to be on KPCC.

When I've checked, the sound has always been the same on the analog or digital signal.

Brian

 

[davideduardo]

The audio samples I posted are from KPCC in June 2015. They are the ones I sent to Lance Harper. Here is his response:

"We're on a new transmitter system as of yesterday and I'm sure I know where in the air chain where the problem is at. It should sound good tomorrow.

The PPM encoder is introducing distortion :-( I'll swap it out tonight."

KPCC's signal did clear up the next day. However, since I had heard the sound for at least two years, I didn't see how it could be due to a "new transmitter system." I also found an e-mail that I sent to Lance in March about the sound appearing on KUOR 89.1, KPCC's outlet in the Inland Empire. It was there for a day and then disappeared. I describe it as the same sound I heard on KPCC. I have not heard it since on KUOR. I rarely can hear KPCC these days so I don't know how it currently sounds.

The sound varies greatly with program material. It is far less audible on music than voice. Although its level differs, the sound character is very similar among the stations I've heard it on. Yesterday I tuned around a bit above 92 MHz. I think I'm hearing the sound on KIIS 102.7, KOST 103.5, and KBIG 104.3. I need to listen more to be sure. It is not as obtrusive on these stations as on KKJZ or KCRW, or as it used to be on KPCC.

When I've checked, the sound has always been the same on the analog or digital signal.

Brian

The obvious question is whether your observations are all made using the same radio receiver. If you try / have tried other radios at different locations to the same result, it is a station issue. Otherwise, it may be a problem with your receiver.

 

[Kelly A]

After listening to the sound again, it sure sounds like a sample rate is set incorrectly in the mic processing or the program chain. Could be just about anywhere where you can set the sample rate in or out of the device.

 

[sdwulfdawg]

A cranked-up Voltair is what I suspect, Kelly. But I don't understand why a public station would sacrifice audio quality in an attempt to boost ratings. I suppose they need ratings to attract underwriters, but I don't believe that is a major source of funding for most public stations.

I heard the funny sound on KPCC for over two years. That may have included time prior to the availability of Voltair. I kept e-mailing Lance Harper about the issue, but the sound persisted. Finally, I e-mailed a recording of some particularly egregious instances and the problem went away the next day. Perhaps I should e-mail a recording to KKJZ. The PPM signal, if that's what I'm hearing, is so strong that it is even audible on breath noise and background studio noise. It's much less obtrusive during music, but once your ear becomes attuned to the characteristic PPM twanginess, you can hear it on just about any source material. I have heard music on KKJZ fade away into a background of pure, continuous twang.

Here are two brief samples of the KPCC sound I recorded in June:

https://www.dropbox.com/s/7g5wd5sbeae46ae/ppm1.wav?dl=1
https://www.dropbox.com/s/om30cc3ngxn0yjx/ppm2.wav?dl=1



Brian

Can you please repost the files? I really wanted to hear them.

 

[k6sti]

KPCC PPM samples recorded in June, 2015:

https://www.dropbox.com/s/wqn3nsp7l7ln22j/kpcc1.wav?dl=1
https://www.dropbox.com/s/0fsz0m5nuxgenyd/kpcc2.wav?dl=1
https://www.dropbox.com/s/ihlez6tw8qzwhcj/kpcc3.wav?dl=1

Response of Steve Herbert, KCRW chief engineer, to my report of excessive PPM levels:

"I appreciate your feed back concerning our PPM encoding. Unfortunately I'm not at liberty to discuss our encoding efforts." The KCRW PPM artifact level seems to vary from day to day. Perhaps they are still tweaking things.

No response from Saul Levine or his engineer Dan Feely about excessive PPM levels on KKJZ, whose signal and programming Mt Wilson FM Broadcasters is responsible for. I've listened to KKJZ for many years, but lately I find the PPM artifacts so annoying that I've removed the preset from all of my radios and tuners. I've switched to KSDS, which has no PPM artifacts or playlist, and has much better jazz.

I'm considering writing code for a PPM cleaner. This real-time audio processor would detect PPM tones and remove them. I would publish the algorithm in the public domain as well as implement it as a free utility for Windows computers. It would use sound card input and output.

Brian

 

[davideduardo]

I'm considering writing code for a PPM cleaner. This real-time audio processor would detect PPM tones and remove them. I would publish the algorithm in the public domain as well as implement it as a free utility for Windows computers. It would use sound card input and output.

Brian

The PPM tags are not tones, but data bursts. They can occur at any of, IIRC, 10 different spots in the audio spectrum where the encoder finds dense enough audio to mask the data burst. The tags are about 4.5 seconds long, and broadcast as often as 12 times a minute if there is masking material.

To remove the tags, you'd have to dynamically remove the actual masking audio, too. In other words, you would be punching holes in the desirable audio while at the same time removing the tag

I would tend to agree with Kelly that the likely culprit is in the actual audio processing of the station, whether it be dueling codecs or a bitrate incompatibility.

 

[k6sti]

David, semantics aside, the data is transmitted as tone bursts. See the following paper and cited patents:

https://www.telosalliance.com/images/25-Seven/Voltair/25-Seven_Tech_Paper_Arbitron_PPM.pdf

To remove the PPM signal, all you need to remove are the added tone bursts. You leave the rest of the audio intact, whether it's involved in masking or not. The simplest method of tone removal may be to cancel the burst by adding an antiphase waveform. This is like using a notch filter, but simpler to implement since detection yields the required tone amplitude and phase. Detection and removal are simple in principle, but there are many details, such as robust synchronization, generation of reliable detection thresholds, and provision for detection errors, that may be tricky to implement. I'm not sure I want to invest the time necessary to do it right. I also wonder how many people could or would use such a utility. For example, I have a PC next to only one radio I use.

You might find the follow paper and audio samples of interest:

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

I checked KKJZ and KCRW this morning. The PPM artifacts are much lower than I recall hearing a few weeks ago. In particular, ambient noise and breath noise no longer generate a continuous stream of artifacts. I wonder if they have dialed back the Voltair enhance level or otherwise altered the PPM injection level. Here's a sample from KKJZ:

https://www.dropbox.com/s/0naixw0vkfyfs1s/kkjz1.wav?dl=1

Brian

 

[davideduardo]

David, semantics aside, the data is transmitted as tone bursts. See the following paper and cited patents:

https://www.telosalliance.com/images/25-Seven/Voltair/25-Seven_Tech_Paper_Arbitron_PPM.pdf

To remove the PPM signal, all you need to remove are the added tone bursts. You leave the rest of the audio intact, whether it's involved in masking or not. The simplest method of tone removal may be to cancel the burst by adding an antiphase waveform. This is like using a notch filter, but simpler to implement since detection yields the required tone amplitude and phase. Detection and removal are simple in principle, but there are many details, such as robust synchronization, generation of reliable detection thresholds, and provision for detection errors, that may be tricky to implement. I'm not sure I want to invest the time necessary to do it right. I also wonder how many people could or would use such a utility. For example, I have a PC next to only one radio I use.

You might find the follow paper and audio samples of interest:

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

I checked KKJZ and KCRW this morning. The PPM artifacts are much lower than I recall hearing a few weeks ago. In particular, ambient noise and breath noise no longer generate a continuous stream of artifacts. I wonder if they have dialed back the Voltair enhance level or otherwise altered the PPM injection level. Here's a sample from KKJZ:

https://www.dropbox.com/s/0naixw0vkfyfs1s/kkjz1.wav?dl=1

Brian

I did not think either of those stations was using a Voltair. With the new encoding enhancements by Nielsen that went into effect in December Nielsen has recommended to all stations that were using them to remove them for all purposes other than metering / monitoring. In theory, no station is using the Voltair for encoding enhancement as there is the very real possibility that doing so could now be considered under the terms of "ratings distortion and bias".

The Voltair was / is a very expensive piece of gear at $15,000 each. The effectiveness was found to be mostly in listening in very noisy environments, and the new Nielsen upgrade in encoding seems superior at handling that situation (ambient masking) than the Voltair was.

Whatever you want to call the data burst (which contains the time and date stamp plus the unique station code) it can occur on any of multiple frequencies as often as 12 to 14 times a minute. In theory, every tag can be on a different frequency. If you notch the frequency, you are also notching not only the tag but the audio at whatever frequency the tag occurs in.

 

[k6sti]

If you notch the frequency, you are also notching not only the tag but the audio at whatever frequency the tag occurs in.

True, but the audio loss should be imperceptible if the PPM algorithm is working properly since it will inject tones only when it thinks the spectrum is masked by nearby audio. Note that for the portable meter to reliably detect a symbol, the residual program audio at the four frequency possibilities must be low enough not to interfere. One trick I had in mind was to use a coefficient less than 1 for the cancelling tone, which will leave some power behind. The coefficient magnitude could reflect detection uncertainty as well as estimated non-PPM power. The real problem I see is that the portable meter knows the masking criteria and can estimate when to expect a symbol. Unless I can figure out the masking criteria, which is a tall order, I'd have only an approximate idea. That could lead to false detection where I cancel spectrum that does not include PPM tones. Even though the notches are very narrow, this may generate artifacts.

I wonder if the improvement I've noticed on KKJZ and KCRW recently is due to their Voltairs being retired with the advent of the updated PPM algorithm.

I forgot to mention that audience applause is the most reliable trigger for audible PPM. I heard it this morning on KPBS, a station on which I've never before heard PPM artifacts. The same thing happened a couple weeks ago on KUOR. The applause acquires an unnatural tonality that is quite distinctive and easily perceived. You can hear the character of the tonality change a couple times per second.

Here is a sample from KKJZ I just recorded that exhibits some breath noise tonality:

https://www.dropbox.com/s/i42n0hzq8n03og9/kkjz2.wav?dl=1

Brian

 

[davideduardo]

I wonder if the improvement I've noticed on KKJZ and KCRW recently is due to their Voltairs being retired with the advent of the updated PPM algorithm.

The first question is whether in fact those two stations even had Voltair units. More FMs in LA did not have them ever than did have them.

The Enhanced CBET firmware was flashed in the largest PPM markets like LA in early December. Those stations using Voltair unints, principally CBS and Clear Channel, were told they should, at most, use them as monitoring devices while Nielsen produces and distributes its own enhance monitoring-capable unit.

The PPM in theory depends on being able to inject the tag into one of four frequencies in one of ten channels. "The PPM system constructs 10 spectral channels in the region from 1.0 kHz to 3.0 kHz. The original program audio energy in each channel is evaluated for its ability to mask an added component. If that masking energy is insufficient, nothing is added." Only if there is enough spectral energy in one of the channels is one of the possible 4 frequencies used to inject the tag. What the Voltair did was increase the spectral energy in the appropriate channel facilitating the insertion of a tag when one might not have been inserted in the original source material. To do that aggressively, artifacts were often noted caused by the use of considerable gain in the channel and near the frequencies of the Nielsen tag.

The MRC has not spoken on this in the framework of the Enghanced CBET upgrade. Any station not following the Nielsen "recommendation" to extinguish the enhancing function of a Voltair could, under an eventual MRC ruling, be penalized for engaging in ratings distortion and bias.

 

[k6sti]

David, do you know if the updated PPM algorithm contains any provision for adjustment?

Brian

 

[TheBigA]

I'm not aware of any non-commercial station that is using Voltair, and I'm pretty sure Saul Levine doesn't use it on his stations.

 

[davideduardo]

David, do you know if the updated PPM algorithm contains any provision for adjustment?

Brian

No. It is firmware flashed to the "old" boxes and there are no controls of any kind. That would defeat the objectives of a level playing field for measurement.