Omnia 11- first look

[BabyDJ]

I'm curious as to what type of equipment was at the demonstration. Screenshots? I don't want to use the "J" word when it comes to this science, but I'm wondering why, if this is such a great idea, it's taken 13 years to "develop" it. With all the advancements in audio processing since 1997, with the addition of HD delays and BS412 limiters, I would think that if this was a legit improvement, someone (especially like Jim Wood) would have done something about it.

Sometimes I think if one of these companies would tell you to mount the box upside down in the rack because it sounds better, we'd have all these upside down processors at transmitter sites.

 

[cgould]

I'm curious as to what type of equipment was at the demonstration. Screenshots? I don't want to use the "J" word when it comes to this science, but I'm wondering why, if this is such a great idea, it's taken 13 years to "develop" it. With all the advancements in audio processing since 1997, with the addition of HD delays and BS412 limiters, I would think that if this was a legit improvement, someone (especially like Jim Wood) would have done something about it.

Because this is something that is VERY difficult to do in analog in light of the heavy amounts of heavy clipping employed in modern day audio processsing without causing severe overshoots. We've been working at this for some time in the background in our lab studying the causes of the overshoots, and using the modern DSP tools in our hands to pull this off with very few side effects (I've decided to be conservative in that statement ).

The processing power we have in our hands these days also allows us to make this a reality at this point in history.

-Cornelius

 

[BabyDJ]

Because this is something that is VERY difficult to do in analog in light of the heavy amounts of heavy clipping employed in modern day audio processsing without causing severe overshoots. We've been working at this for some time in the background in our lab studying the causes of the overshoots, and using the modern DSP tools in our hands to pull this off with very few side effects (I've decided to be conservative in that statement ).

The processing power we have in our hands these days also allows us to make this a reality at this point in history.

-Cornelius

I dare say that I think there was enough horsepower in the 6 to do this much earlier, if you consider the 6 started out as basically an FM only processor. Since its inception, it has seen the addition of HD delay and a totally separate path for digital.

I think some of what Bob Orban says needs to be taken into consideration before people start to go nuts on this.

 

[rorban]

I think some of what Bob Orban says needs to be taken into consideration before people start to go nuts on this.

I am still waiting for an answer to my question about the low frequency cutoff of the SSB stereo subchannel in Frank's realization because this will determine if highly processed material will produce overshoots in the composite waveform that must be removed by nonlinear means if current US modulation rules are observed.

The VSB technique that I proposed does not have this disadvantage because the frequency response of the stereo subchannel can then extend to DC (the way it does in the current FM stereo system) and the technique will not add overshoots to highly processed material.

Just to be clear, DC response is the ideal way of preserving the shapes of highly processed waveforms. It does not imply that there is a DC component in the audio, just that highly processed audio typically introduces a small amount of energy at very low frequencies that was not in the original program material and it is necessary to pass this energy through the transmission system without distorting its waveform. The 0.15 Hz limit could probably be raised in a fully phase-linear system because there is no nonlinear group delay to tilt low frequency squarewaves. I don't know offhand what the necessary system bandwidth would be to constrain overshoots to 1% or less, but the high frequency cutoff would definitely be have to be lower than 16 Hz to accommodate all known program material. (Some pipe organs produce frequencies this low and a system's technical specifications should allow it to be transparent even to niche program material.) Achieving even 16 Hz in a fully SSB system would require long filters, while the VSB technique is a fully linear solution to the problem of peak modulation control that removes the need to compromise between peak modulation control, coding delay, and DSP load. Instead, coding delay and DSP load would be traded off against the bandwidth of the upper sideband.

Another possible approach to the problem is to do a low frequency blend before the processor's peak control elements, redirecting bass energy from the L-R to the L+R , which is something that some broadcasters already do to reduce multipath distortion and which is a feature that we plan to add to the 8600 soon. This would minimize the low frequency energy in the L-R. However, the peak control elements, because they are handling non-identical program material, will typically still produce a small amount of L-R energy when operated to achieve what is consider to be "competitive loudness." It is conceivable that this could cause the composite to overshoot. Moreover, bass blend requires compromising the integrity of the original program material and is definitely not something that is appropriate to build into a technical specification for an improved FM stereo modulation technique. Instead, choosing bass blend should up to individual broadcasters.

Bob Orban

 

[FFoti1]

I dare say that I think there was enough horsepower in the 6 to do this much earlier, if you consider the 6 started out as basically an FM only processor. Since its inception, it has seen the addition of HD delay and a totally separate path for digital.

I think some of what Bob Orban says needs to be taken into consideration before people start to go nuts on this.

If there was enough power left in Omnia.6 to have done this, we would have. Moore's Law applies.

BTW: When we added diversity delay, it required a different DSP card that contained the required memory for the delay function. We gave the SSB idea consideration, at that time, but again were limited to resources that were available.

I need to point something out regarding this idea. This isn't an Omnia vs the other guys situation. It's based on seeing if the concept will benefit broadcasting. If it does, I'm sure the feature will become available in your stereo generator de jour. How about if we look at this, as an industry, to see if there are merits in it.

-Frank Foti

 

[FFoti1]

I am still waiting for an answer to my question about the low frequency cutoff of the SSB stereo subchannel in Frank's realization because this will determine if highly processed material will produce overshoots in the composite waveform that must be removed by nonlinear means if current US modulation rules are observed.

Bob Orban

We researched the issue regarding the SSB L-R overshoot topic, and came up with a filter implementation that yields extremely low overshoot under heavy processing conditions. The overshoot content is no more than observed on your 8500, with the composite clipper turned off, when the same type of processing is employed.

-Frank Foti

 

[FFoti1]

I'm curious as to what type of equipment was at the demonstration. Screenshots? I don't want to use the "J" word when it comes to this science, but I'm wondering why, if this is such a great idea, it's taken 13 years to "develop" it.

This didn't take 13 years to develop. I've had Bill Gillman's paper that long. I began working on this during a break after Christmas 2009, and had it running by NAB 2010.

Equipment is all of the garden radio station variety, along with the prototype SSB stereo generator.

-Frank Foti

 

[rorban]

I am still waiting for an answer to my question about the low frequency cutoff of the SSB stereo subchannel in Frank's realization because this will determine if highly processed material will produce overshoots in the composite waveform that must be removed by nonlinear means if current US modulation rules are observed.

Bob Orban

We researched the issue regarding the SSB L-R overshoot topic, and came up with a filter implementation that yields extremely low overshoot under heavy processing conditions. The overshoot content is no more than observed on your 8500, with the composite clipper turned off, when the same type of processing is employed.

-Frank Foti

Frank -- Is this filter nonlinear? Does it add more spectrum when it operates?

You still haven't answered the question regarding the low frequency cutoff of your SSB implementation. If you are serious about making a proposal to the NRSC and FCC regarding a variation of the FM stereo composite waveform, you need to be more forthcoming about these details. If it requires nonlinear filtering for modulation control, then my VSB proposal is preferable because it requires no nonlinear processing to achieve response to DC and perfect composite modulation control (not counting the pilot tone) provided only that the left and right inputs are correctly peak-controlled.

In the current FM stereo system, the peak modulation is the higher of the left and right waveforms before the pilot is added. Because the pilot is correlated to the suppressed subcarrier, it turns out that for 9% pilot injection, pure L+R produces 2.7% higher composite modulation (+0.23 dB) than pure left or right modulation at the same peak level applied to the stereo encoder. This 0.23 dB is the maximum modulation penalty that one pays in the current DSB system. The loudness difference is right on the threshold of audibility.

I have not yet examined pilot interleaving in the SSB system; it's on my to-do list.

Bob Orban

 

[FFoti1]

Frank -- Is this filter nonlinear? Does it add more spectrum when it operates?

You still haven't answered the question regarding the low frequency cutoff of your SSB implementation. If you are serious about making a proposal to the NRSC and FCC regarding a variation of the FM stereo composite waveform, you need to be more forthcoming about these details. If it requires nonlinear filtering for modulation control, then my VSB proposal is preferable because it requires no nonlinear processing to achieve response to DC and perfect composite modulation control (not counting the pilot tone) provided only that the left and right inputs are correctly peak-controlled.

In the current FM stereo system, the peak modulation is the higher of the left and right waveforms before the pilot is added. Because the pilot is correlated to the suppressed subcarrier, it turns out that for 9% pilot injection, pure L+R produces 2.7% higher composite modulation (+0.23 dB) than pure left or right modulation at the same peak level applied to the stereo encoder. This 0.23 dB is the maximum modulation penalty that one pays in the current DSB system. The loudness difference is right on the threshold of audibility.

I have not yet examined pilot interleaving in the SSB system; it's on my to-do list.

Bob Orban

Bob,

We're extremely serious about this SSB idea. So much so, I inquired with a number of key engineers in our industry about the feasibility of the it. Most whom you know, and I'd venture to say, you respect. One of them, even had an informal chat with the 'Boys in Washington,' who felt the idea was worth looking into further. So, we have. The NRSC asked me to give an overview, of the idea at the NAB Radio Show, in Washington. After the preso, they asked me to do some in-field testing, which will begin soon. This is not some hobbyist project Frank took on for the sake of it. (The live steam engine keeps me pretty busy in that regard!) As stated before, the intent is for the benefit of our medium. One that's suffered, due to various reasons (we need not go into those here), and if we can offer broadcasters something that will benefit their transmissions, then why are we debating over cut-off frequencies? Seems we're from a fraternity of smart people, and here's a chance to do something that quite possibly may offer benefit to the end-user...the listener!

Now, regarding your question about our application. You're not the first to think about using VSB for the SSB filter. We've implemented it and it works, just as you say. Also, I came up with an innovative method that does not require a long delay, nor does it generate added spectra. We intend to field test both methods. I have composite peak control in SSB that is the same as it would be in DSB, with no added spectra.

If you wish to join in the fun, by all means, let's band together and see what's possible.

-Frank Foti

 

[BabyDJ]

Has it actually been onthe air in a real world situation... like San Francisco for starters. I know Bob can verify that challenge

 

[rorban]

We're extremely serious about this SSB idea. So much so, I inquired with a number of key engineers in our industry about the feasibility of the it. Most whom you know, and I'd venture to say, you respect. One of them, even had an informal chat with the 'Boys in Washington,' who felt the idea was worth looking into further. So, we have. The NRSC asked me to give an overview, of the idea at the NAB Radio Show, in Washington. After the preso, they asked me to do some in-field testing, which will begin soon. This is not some hobbyist project Frank took on for the sake of it. (The live steam engine keeps me pretty busy in that regard!) As stated before, the intent is for the benefit of our medium. One that's suffered, due to various reasons (we need not go into those here), and if we can offer broadcasters something that will benefit their transmissions, then why are we debating over cut-off frequencies? Seems we're from a fraternity of smart people, and here's a chance to do something that quite possibly may offer benefit to the end-user...the listener!

Now, regarding your question about our application. You're not the first to think about using VSB for the SSB filter. We've implemented it and it works, just as you say. Also, I came up with an innovative method that does not require a long delay, nor does it generate added spectra. We intend to field test both methods. I have composite peak control in SSB that is the same as it would be in DSB, with no added spectra.

If you wish to join in the fun, by all means, let's band together and see what's possible.

-Frank Foti

Frank,

You did not answer my two questions:

1. Is your peak control method nonlinear? That is, do scaling and superposition hold or do they not? By "adding spectra," I mainly meant "does it add *in-band* spectrum" (that would decode in a standard FM stereo radio as nonlinear distortion). I presumed that your technique technique would not add out-of-band spectrum any more than today's band-limited composite clipping or limiting techniques would. However, today's composite limiters or clippers tend to add in-band spectrum that decodes in the radio as nonlinear distortion even while they adequately protect the pilot tone and the baseband spectrum above about 55 kHz.

2. Over what frequency range in the stereo subchannel is your proposed system truly SSB? (Why do you keep dodging this question?) The "debate" is over whether VSB (which is 100% linear) or your proprietary technique is likely to be preferable, but you have not revealed enough about your technique to even allow us to debate this in an informed manner. If you are going to propose a rulemaking that changes the current FM stereo rules, you are going to have to reveal what the technical details of your proposal actually are. I *do* believe that if the existing DSB system (which is ideally 100% linear) is to be changed, it is highly desirable that any substitute also be linear, particularly because VSB with a practical SSB/VSB crossover frequency can retain full linearity in an elegant way while adding less than 1% to the baseband occupied bandwidth compared to a pure SSB system. A further advantage of VSB is that it is non-proprietary and well understood.

I am in favor of any changes that will in fact benefit the listener. But when one proposes changing FCC and ITU-R rules, regulations, and standards, it is necessary to supply a great deal of data. The proposed system must be completely described technically (like, for example, the ATSC digital television system and *not* like the HD Radio system with its black-box codec) and must be thoroughly tested to determine receiver compatibility and co-channel RF protection ratios. (In my opinion, based on the RF spectra pictures in your article, adjacent channel RF protection ratios should not be a problem because of the system's lower bandwidth.)

I am somewhat concerned about receiver compatibility because I do not know how many radios out there use the quadrature component of the demodulated stereo subchannel to estimate transmission channel degradations that would demand triggering variable blend-to-mono. Maybe the Sony is the only one. But guessing is not good enough. Broadcasters need to know if implementing the new system will force a significant number of the listeners' radios into blending, which can also include HF rolloff.

I think that this is an excellent project for the NRSC and I would certainly be interested being involved with a working group to investigate this. I suspect that NPR Labs probably has the necessary equipment, particularly the multipath simulator and other aspects of the RF path instrumentation.

Bob Orban

 

[FFoti1]

Frank,

You did not answer my two questions:

1. Is your peak control method nonlinear? That is, do scaling and superposition hold or do they not? By "adding spectra," I mainly meant "does it add *in-band* spectrum" (that would decode in a standard FM stereo radio as nonlinear distortion). I presumed that your technique technique would not add out-of-band spectrum any more than today's band-limited composite clipping or limiting techniques would. However, today's composite limiters or clippers tend to add in-band spectrum that decodes in the radio as nonlinear distortion even while they adequately protect the pilot tone and the baseband spectrum above about 55 kHz.

2. Over what frequency range in the stereo subchannel is your proposed system truly SSB? (Why do you keep dodging this question?) The "debate" is over whether VSB (which is 100% linear) or your proprietary technique is likely to be preferable, but you have not revealed enough about your technique to even allow us to debate this in an informed manner. If you are going to propose a rulemaking that changes the current FM stereo rules, you are going to have to reveal what the technical details of your proposal actually are. I *do* believe that if the existing DSB system (which is ideally 100% linear) is to be changed, it is highly desirable that any substitute also be linear, particularly because VSB with a practical SSB/VSB crossover frequency can retain full linearity in an elegant way while adding less than 1% to the baseband occupied bandwidth compared to a pure SSB system. A further advantage of VSB is that it is non-proprietary and well understood.

I am in favor of any changes that will in fact benefit the listener. But when one proposes changing FCC and ITU-R rules, regulations, and standards, it is necessary to supply a great deal of data. The proposed system must be completely described technically (like, for example, the ATSC digital television system and *not* like the HD Radio system with its black-box codec) and must be thoroughly tested to determine receiver compatibility and co-channel RF protection ratios. (In my opinion, based on the RF spectra pictures in your article, adjacent channel RF protection ratios should not be a problem because of the system's lower bandwidth.)

I am somewhat concerned about receiver compatibility because I do not know how many radios out there use the quadrature component of the demodulated stereo subchannel to estimate transmission channel degradations that would demand triggering variable blend-to-mono. Maybe the Sony is the only one. But guessing is not good enough. Broadcasters need to know if implementing the new system will force a significant number of the listeners' radios into blending, which can also include HF rolloff.

I think that this is an excellent project for the NRSC and I would certainly be interested being involved with a working group to investigate this. I suspect that NPR Labs probably has the necessary equipment, particularly the multipath simulator and other aspects of the RF path instrumentation.

Bob Orban

Bob,

If you look at the paper, you'll see the SSB range is indicated. It has been implemented using various SSB filters where we can set the range wherever we need it. The answer to your question is we can set this extremely low, practically down to 38kHz. At this time, we have multiple iterations of the design, which will be tested, likewise the same for the peak control method. Once we settle on a finalized implementation, revelations will be made, at that time. I sense your quest to know the exactness of what we're doing and why, and once more testing is done, you will receive your answer.

Since, none of this has reached the in-depth NRSC, or FCC evaluation level yet, I see no need to put specifics out there. We're about to go into field testing, and that will provide plenty of information as to whether or not the idea is carried further.

You wonder about receivers? Well, we've got receiver companies lined up to do testing, and I've already had a chat with the NPR labs as well.

Again, if you'd like to be involved, we welcome your participation. If not, sit tight and your questions will be answered. I understand something of this nature needs testing, evaluation, and disclosure. As previously stated, the regulating authorities are aware of our efforts, and have encouraged us to carry on with more testing. Results of those test results will be evaluated to determine, if we take it to the FCC and ITU-R. At that time, all info will be shared.

-Frank Foti