Reply by Jerry Avins February 9, 20062006-02-09
Mark wrote:
> the carrier going down and sidebands fading differentially CAN be > corrected by an adaptive equalizer (not the same thing as an audio guys > call a graphic equlaizer) that trys to keep the channel flat...this is > done all the time with QAM signals..... but my question was how do you > expect to be able to do this on a voice modulated carrier AM signal > that has no "expected" characteristics...For example, in the simplest > case, when there is no modulation, you have only carrier so there is no > way an AE can have any information about selective fading in the > channel.... > > OK maybe if some sort of low level pre-defined reference signal was > sent by the Tx, but this is not done...
In practice, one doesn't make it good, one settles for making it better. There's a lot of redundancy in voice that the brain can use if overall volume is held fairly level and distortion is avoided. Early exalted-carrier receivers had a sharp, narrow peak in the center of the IF, so that the carrier amplitude at the detector was 20 dB over 100% modulated sideband; this prevented most distortion from fading carrier. Modern practice achieves the same result by injecting a phase-locked oscillator. Various AGC schemes are used in the IF strip, but final AGC is done on the demodulated audio. The scheme wouldn't do for audiophiles, but it works for mobile dispatch. For the base station, diversity antennas provide even more improvement. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by Mark February 8, 20062006-02-08
the carrier going down and sidebands fading differentially CAN be
corrected by an adaptive equalizer (not the same thing as an audio guys
call a graphic equlaizer)  that trys to keep the channel flat...this is
done all the time with QAM signals..... but my question was how do you
expect to be able to do this on a voice modulated carrier AM signal
that has no "expected" characteristics...For example, in the simplest
case, when there is no modulation, you have only carrier so there is no
way an AE can have any information about selective fading in the
channel....

OK maybe if some sort of low level pre-defined reference signal was
sent by the  Tx, but this is not done...
Mark

Reply by Jerry Avins February 8, 20062006-02-08
Anonymous wrote:
> "Jerry Avins" <jya@ieee.org> wrote in message > news:1I2dndx7iPRC73fenZ2dnUVZ_tudnZ2d@rcn.net... > >>Anonymous wrote: >> >>>"Vladimir Vassilevsky" <antispam_bogus@hotmail.com> wrote in message >>>news:A1uGf.54199$PL5.3648@newssvr11.news.prodigy.com... >>> >>> >>>>Anonymous wrote: >>>> >>>> >>>> >>>>>I basically agree but you can minimize the fatigue on the listener by >>>>>getting rid of the noise bursts that occur as the fade drives the > > signal > >>>>>towards the noise floor, right? >>>> >>>>What you are trying to build is not the equalizer but the adaptive noise >>>>reduction. Yes you can get some perceptual improvement, however the >>>>sound will suffer "unnaturally deep" , "watery" and "rubber like" >>> >>>artifacts. >>> >>> >>>>Vladimir Vassilevsky >>>> >>>>DSP and Mixed-Up Signal Design Consultant >>>> >>>>http://www.abvolt.com >>> >>> >>> >>>So what do you think on an AM voice signal can be equalized? >> >>I can only begin to answer that if you explain what you want your >>equalizer to do. Volume? Frequency response? Phase distortion? >> >>Please tell me what you want to do. The name you give the gizmo that >>does it is unimportant. >> >>Jerry >>-- >>Engineering is the art of making what you want from things you can get. >>&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095; > > > I believe the original problem was modulation on his AM voice signal, i.e. > fading was causing the volume at the speaker to waver.
The OP called it an equalizer; it's his explanation I want. AGC does a pretty good job of keeping the carrier level constant. Most of the volume change is caused by sidebands fading differently from the carrier. If the carrier only fades, the volume goes up and the signal gets distorted. There's no good way around that for music, but there are known methods for voice communications. Exalted carrier and audio AGC is a good way, but I wouldn't call that an equalizer. Jerry -- Engineering is the art of making what you want from things you can get. &#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;
Reply by Anonymous February 8, 20062006-02-08
"Jerry Avins" <jya@ieee.org> wrote in message
news:1I2dndx7iPRC73fenZ2dnUVZ_tudnZ2d@rcn.net...
> Anonymous wrote: > > "Vladimir Vassilevsky" <antispam_bogus@hotmail.com> wrote in message > > news:A1uGf.54199$PL5.3648@newssvr11.news.prodigy.com... > > > >> > >>Anonymous wrote: > >> > >> > >>>I basically agree but you can minimize the fatigue on the listener by > >>>getting rid of the noise bursts that occur as the fade drives the
signal
> >>>towards the noise floor, right? > >> > >>What you are trying to build is not the equalizer but the adaptive noise > >>reduction. Yes you can get some perceptual improvement, however the > >>sound will suffer "unnaturally deep" , "watery" and "rubber like" > > > > artifacts. > > > >>Vladimir Vassilevsky > >> > >>DSP and Mixed-Up Signal Design Consultant > >> > >>http://www.abvolt.com > > > > > > > > So what do you think on an AM voice signal can be equalized? > > I can only begin to answer that if you explain what you want your > equalizer to do. Volume? Frequency response? Phase distortion? > > Please tell me what you want to do. The name you give the gizmo that > does it is unimportant. > > Jerry > -- > Engineering is the art of making what you want from things you can get. > &#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;
I believe the original problem was modulation on his AM voice signal, i.e. fading was causing the volume at the speaker to waver.
Reply by Jerry Avins February 8, 20062006-02-08
Anonymous wrote:
> "Vladimir Vassilevsky" <antispam_bogus@hotmail.com> wrote in message > news:A1uGf.54199$PL5.3648@newssvr11.news.prodigy.com... > >> >>Anonymous wrote: >> >> >>>I basically agree but you can minimize the fatigue on the listener by >>>getting rid of the noise bursts that occur as the fade drives the signal >>>towards the noise floor, right? >> >>What you are trying to build is not the equalizer but the adaptive noise >>reduction. Yes you can get some perceptual improvement, however the >>sound will suffer "unnaturally deep" , "watery" and "rubber like" > > artifacts. > >>Vladimir Vassilevsky >> >>DSP and Mixed-Up Signal Design Consultant >> >>http://www.abvolt.com > > > > So what do you think on an AM voice signal can be equalized?
I can only begin to answer that if you explain what you want your equalizer to do. Volume? Frequency response? Phase distortion? Please tell me what you want to do. The name you give the gizmo that does it is unimportant. Jerry -- Engineering is the art of making what you want from things you can get. &#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;
Reply by Anonymous February 8, 20062006-02-08
"Vladimir Vassilevsky" <antispam_bogus@hotmail.com> wrote in message
news:A1uGf.54199$PL5.3648@newssvr11.news.prodigy.com...
> > > Anonymous wrote: > > > > > I basically agree but you can minimize the fatigue on the listener by > > getting rid of the noise bursts that occur as the fade drives the signal > > towards the noise floor, right? > > > > > What you are trying to build is not the equalizer but the adaptive noise > reduction. Yes you can get some perceptual improvement, however the > sound will suffer "unnaturally deep" , "watery" and "rubber like"
artifacts.
> > Vladimir Vassilevsky > > DSP and Mixed-Up Signal Design Consultant > > http://www.abvolt.com
So what do you think on an AM voice signal can be equalized?
Reply by Vladimir Vassilevsky February 8, 20062006-02-08

Anonymous wrote:

> > I basically agree but you can minimize the fatigue on the listener by > getting rid of the noise bursts that occur as the fade drives the signal > towards the noise floor, right?
>
What you are trying to build is not the equalizer but the adaptive noise reduction. Yes you can get some perceptual improvement, however the sound will suffer "unnaturally deep" , "watery" and "rubber like" artifacts. Vladimir Vassilevsky DSP and Mixed-Up Signal Design Consultant http://www.abvolt.com
Reply by Anonymous February 8, 20062006-02-08
"Vladimir Vassilevsky" <antispam_bogus@hotmail.com> wrote in message
news:jYpGf.53436$PL5.14607@newssvr11.news.prodigy.com...
> > > Anonymous wrote: > > > > > > The basical principle is you use the adaptive filter to "predict" the
actual
> > received signal, i.e. the error signal driving the AF is the difference > > between the actual signal and the output of the AF. Then if you listen
to
> > the output of the AF you get only the components that are statistically > > predictable. In my case I used the AF output as the speech enhancement > > output (you can't predict noise or rapid fading, right?) and I used the > > error signal for tone removal. By tweaking the step size of the AF you
can
> > control the statistics it's able to track from highly correlated(tones)
to
> > medium correlated(speech). > > Are you implementing some kind of noise reduction filter after the AM > envelope detector? Or are you trying to compensate for the multipath > distortion before the detector? What is the goal? >
The former. A fast AGC on the IF could compensate the fading but it can't be so fast that it strips the AM modulation, right?
> > The advantage of the AF is that it automatically expands/contracts the > > bandwidth based on the SNR of the received signal. There are several HF > > products out there doing the same thing by the way. > > There is not much bandwidth in AM. It's only about 5kHz. Since the > spectrum is so narrow, the fading is not very selective and hits a big > part of the spectrum at once. I don't expect much gain from either > pre-detector or post-detector processing. >
I basically agree but you can minimize the fatigue on the listener by getting rid of the noise bursts that occur as the fade drives the signal towards the noise floor, right?
> Vladimir Vassilevsky > > DSP and Mixed Signal Design Consultant > > http://www.abvolt.com
Reply by Anonymous February 8, 20062006-02-08
"Mark" <makolber@yahoo.com> wrote in message
news:1139417980.646987.93170@g44g2000cwa.googlegroups.com...
> Clark, > > The AF (adaptive filter) you are describing automatically adapts the > overall bandwidth of the signal...OK...good... but does it actually > compensate for selective fading the way an adaptive equalizer does for > digtial modulation... i.e. if the selective fading causes a 5 dB tilt > across the channel, will your AF compensate for that tilt by creating > a reverse tilt or will it just reduce the bandwidth because the SNR is > poor.... > > Mark >
An AF needs some known component or statistic by which to generate an error term from. With voice AM there really isn't any fixed parameter to lean on. A loss of amplitude could be fading on the RF OR the person speaking could have just stopped talking. If it were FM you could probably uses the constant envelope of the signal, but I don't see any way to do it with AM. Maybe ALC on the audio will work for you I just know that an AF in the predictive mode worked well for us in the past to make HF audio much more tolerable to listen to. There used to be several companies selling little dsp boxes that you hung off your radio speaker to do this. I can recall a name right now. -Clark
Reply by Vladimir Vassilevsky February 8, 20062006-02-08

Anonymous wrote:


> > The basical principle is you use the adaptive filter to "predict" the actual > received signal, i.e. the error signal driving the AF is the difference > between the actual signal and the output of the AF. Then if you listen to > the output of the AF you get only the components that are statistically > predictable. In my case I used the AF output as the speech enhancement > output (you can't predict noise or rapid fading, right?) and I used the > error signal for tone removal. By tweaking the step size of the AF you can > control the statistics it's able to track from highly correlated(tones) to > medium correlated(speech).
Are you implementing some kind of noise reduction filter after the AM envelope detector? Or are you trying to compensate for the multipath distortion before the detector? What is the goal?
> The advantage of the AF is that it automatically expands/contracts the > bandwidth based on the SNR of the received signal. There are several HF > products out there doing the same thing by the way.
There is not much bandwidth in AM. It's only about 5kHz. Since the spectrum is so narrow, the fading is not very selective and hits a big part of the spectrum at once. I don't expect much gain from either pre-detector or post-detector processing. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com