Pages: [1] 2 3   Go Down

Author Topic: Audio output DSPs and Parametric EQ speech enhancement  (Read 500 times)

KK4GMU

  • Member
  • Posts: 366
Audio output DSPs and Parametric EQ speech enhancement
« on: July 17, 2022, 02:48:14 PM »

There are several approaches taken by manufacturers to enhance legibility of speech within noisy signals.

Since the human voice is centered at ~2,500 Hz, with speech extending roughly from 500 to 4,000 Hz, this is the range that needs to be emphasized.

Heil prefers the parametric EQ approach, where he uses 2 or 3 channels and enables changing the center frequency, bandwidth and amplitude in each.  Most new radios have something similar built in, often minus the ability to change the center point or the audio bandwidth. His favorite boost frequency is 2,500 Hz.  The frequencies below and above 2,500 are cut.

This is very much like the standard graphic equilizer, but with the focus on frequencies used in ham radio audio.  There is little to no use for frequencies below 500 or frequencies above 4,000.

Several other manufacturers just cut frequencies below 1,000 Hz and above 3,000 Hz which has a similar effect and with a little gain added to the entire frequency band.

Then there are the Digital Signal Processors (DSPs).  This is where it gets fuzzy for me.  These systems sample the signals, sample the noise, and separate out what it believes to be voice and voila, a more legible voice minus most noise.

For those who have used EITHER or BOTH techniques via an external component to enhance voice clarity, how have they worked for you?  Which one does the better job?

(And yes, yes, I understand that eliminating noise BEFORE the audio stage is preferred.  This post RF audio topic is directed toward those of us who have done all we care to do on the RF end and want to accomplish what we can on the output/audio side.)

The external, stand alone products I am considering are these:

XIEGU GBR-1  https://qrznow.com/xiegu-gnr-1-noise-reduction-unit-full-review/

Timewave DSP-599 or DSP-9 (Both discontinued???)

West Mountain Radio CLRdsp    http://www.westmountainradio.com/product_info.php?products_id=clr_dsp

BHI (various models), mainly parametrics   https://www.bhi-ltd.com/

Are there others you have had good experience with?

Please describe the type of noise your solution has best suppressed.
My noise is unidentified neighborhood or atmospheric broadband.  I already shut down everything electrical in my house, run my radio off of a lead acid battery.  Going out 100 feet around my house with an HT I get solid S9 on 2 meters on up. 

So, at this point I am only interested in potential radio audio output processors that enhance voice clarity buried in noise, whether parametric EQ, DSP, or a combination.

Oh, and a few of the demonstrations I have seen on a couple of these products demonstrate achieving better voice clarity than can be achieved with the NR in even newer radios.  But I take those with a grain of salt until I get more user feedback.  I have an Icom IC-7100.  And yes, I've played with and set the internal tone and noise settings.
« Last Edit: July 17, 2022, 03:10:14 PM by KK4GMU »
Logged
IC-7100, RSPdx, AT-D878UVII-Plus HT, TGIF Spot

W6MK

  • Posts: 4095
    • HomeURL
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #1 on: July 17, 2022, 04:12:47 PM »

A. Then there are the Digital Signal Processors (DSPs).  This is where it gets fuzzy for me.  These systems sample the signals, sample the noise, and separate out what it believes to be voice and voila, a more legible voice minus most noise.

B. How have they worked for you?  Which one does the better job?

C. And yes, yes, I understand that eliminating noise BEFORE the audio stage is preferred.  This post RF audio topic is directed toward those of us who have done all we care to do on the RF end.

D. I have seen on a couple of these products demonstrate achieving better voice clarity than can be achieved with the NR in even newer radios.

A. Audio equalizers and DSPs that I've used have minor beneficial effects on "speech quality." All of the ones I've tried have had some effect, but I wouldn't plan on any miraculous benefit.

B. I mostly operate CW. The great benefit of this mode is that by using a very narrow IF bandwidth the signal-to-noise can be improved greatly. This is not possible with SSB. Appropriate use of RF gain can also be very helpful. Use minimal RF gain.

We live in a very noisy RF world. There is no simple solution which involves buying one product or another. Some products may work better under some noise conditions. The only way to compare devices is to try as many as you can. Buy (used), steal or borrow.

"This post RF audio topic is directed toward those of us who have done all we care to do on the RF end." Really?

C. Yes, eliminate the noise as early in the signal chain as you can. Devices which can be most helpful include phasing units and noise antennas, receive antennas and directional antennas.

D. Commercial products may do some good, but try to think critically about online demonstrations and the usual advertising hype.
Logged

K7JQ

  • Member
  • Posts: 2602
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #2 on: July 17, 2022, 04:16:27 PM »

I use the WMR CLRdsp (newest model with continuously variable filter, tone, and volume controls) with both of my IC-7300’s. Very satisfied with them, although you’ll see some reviews that don’t agree with me.

The NR on the 7300 is not bad, not giving you the “watery” sounding artifacts like other radios have had on their NR circuits. But I was not happy that as you increase the NR level control, the volume decreases and the voice quality becomes muffled. CW also seems stifled with increased levels.

Adding the CLRdsp *clearly* reduces the background noise/hiss (up to about 90%) while maintaining the audio quality and volume, even with the Filter level control close to fully clockwise. No watery artifacts or muffled audio. I’m talking about atmospheric band and some man made noise from household RF hash emitting electronics. Not “pulse type” noise, although the 7300 NB does a good job of eliminating that.

The DSP algorithm is an adaptive one, meaning it might take a few seconds for the circuit to sample the noise and eliminate/reduce it. Happening mostly on stronger signals, when going from transmit to receive, you might get a “whoosh” of noise before it settles down. I found that when I engage the 7300’s NR at a low level (#2) in tandem with the CLRdsp, that whoosh is not as noticible and reduced.

All in all, they’re keepers for me. Good luck with your decision.

73, Bob K7JQ

Logged

K6AER

  • Member
  • Posts: 7159
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #3 on: July 17, 2022, 05:03:12 PM »

There are several approaches taken by manufacturers to enhance legibility of speech within noisy signals.

Since the human voice is centered at ~2,500 Hz, with speech extending roughly from 500 to 4,000 Hz, this is the range that needs to be emphasized.

Heil prefers the parametric EQ approach, where he uses 2 or 3 channels and enables changing the center frequency, bandwidth and amplitude in each.  Most new radios have something similar built in, often minus the ability to change the center point or the audio bandwidth. His favorite boost frequency is 2,500 Hz.  The frequencies below and above 2,500 are cut.

This is very much like the standard graphic equilizer, but with the focus on frequencies used in ham radio audio.  There is little to no use for frequencies below 500 or frequencies above 4,000.

Several other manufacturers just cut frequencies below 1,000 Hz and above 3,000 Hz which has a similar effect and with a little gain added to the entire frequency band.

Then there are the Digital Signal Processors (DSPs).  This is where it gets fuzzy for me.  These systems sample the signals, sample the noise, and separate out what it believes to be voice and voila, a more legible voice minus most noise.

For those who have used EITHER or BOTH techniques via an external component to enhance voice clarity, how have they worked for you?  Which one does the better job?

(And yes, yes, I understand that eliminating noise BEFORE the audio stage is preferred.  This post RF audio topic is directed toward those of us who have done all we care to do on the RF end and want to accomplish what we can on the output/audio side.)

The external, stand alone products I am considering are these:

XIEGU GBR-1  https://qrznow.com/xiegu-gnr-1-noise-reduction-unit-full-review/

Timewave DSP-599 or DSP-9 (Both discontinued???)

West Mountain Radio CLRdsp    http://www.westmountainradio.com/product_info.php?products_id=clr_dsp

BHI (various models), mainly parametrics   https://www.bhi-ltd.com/

Are there others you have had good experience with?

Please describe the type of noise your solution has best suppressed.
My noise is unidentified neighborhood or atmospheric broadband.  I already shut down everything electrical in my house, run my radio off of a lead acid battery.  Going out 100 feet around my house with an HT I get solid S9 on 2 meters on up. 

So, at this point I am only interested in potential radio audio output processors that enhance voice clarity buried in noise, whether parametric EQ, DSP, or a combination.

Oh, and a few of the demonstrations I have seen on a couple of these products demonstrate achieving better voice clarity than can be achieved with the NR in even newer radios.  But I take those with a grain of salt until I get more user feedback.  I have an Icom IC-7100.  And yes, I've played with and set the internal tone and noise settings.

A parametric EQ will not do much for the normal SSB radio for the radio IF is only going to pass audio up to 2.7 KHz.

First off the male human speech is mostly 200 Hz to 800Hz, which is the range of speech power/projection  and the articulation range from 1200 Hz to 2400HZ.

You will notice that the band width of a normal SSB signal is rarely more then 2.8 KHz wide.

The best SSB signals keep the levels with about 3 dB more boost on the lows than the highs.

ESSB was all the rage up to about 6 years ago and the only thing is accomplished was wasted band width and adjacent channel interference.

If you have an external noise radiation getting into your antenna pattern you might want to use a QRM eliminator (ANC-4, MFJ1026) with an separate reference antenna. Contrarily to popular belief, they do not reduce interference but place a null in you antenna pattern in the direction of the interference source. If you have multiple interference sources (different directions) this system will not work well.

Bottom line is the recovered audio minus the noise floor is what you have to work with in audio reproduction. Because white noise is a function of all audio frequencies, you are best to center your audio response around the real speech frequencies.

If the ham on the other end has excessively boosted his audio response to sound like some Friday Night DJ, it is at the expense of communication clarity.

« Last Edit: July 17, 2022, 05:06:00 PM by K6AER »
Logged

KM3F

  • Member
  • Posts: 1103
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #4 on: July 17, 2022, 11:08:16 PM »

I have a different opinion on how the ANC 4 works at my setup.
I use a switching supply to power a TS 2000 setting right aside the radio.
I can take the Antenna off the ANC 4 and just with the sample wire close by totally clear the SSP interfering signal pickup at that close separation.
It has nothing to do with Rx pattern nulling.
Inside the ANC 4 both signals are present in opposite polarities as long as the Phase control is adjusted for that cancellation to happen before the result gets to the Rx.
Works the same when the antenna is connected as long as the signal levels are not too high on one of the signals such that complete cancellation cannot be achieved.
This is also why an interfering signal from sky wave or long distances cannot be cancelled as is cannot be sensed by the ANC 4 at a high enough level to do any cancelling.

Logged

KK4GMU

  • Member
  • Posts: 366
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #5 on: July 18, 2022, 04:09:28 AM »

ANC-4 or MFJ-1026 off topic and wrong subject category. That would be the "RFI/EMI" category.   I'd like to stick to post RF solutions. Tried it. In my case no cigar. Thanks.

I viewed a few YouTubes by Bob Heil.  He is big on the parametrics, but not so much on DSP, apparently.  The DSP technology seems to be better suited to noise reduction, although I can see the two in combination - parametric equalization AND DSP noise reduction - being a good combination to squeeze just a bit more clarity from the murk.
 
« Last Edit: July 18, 2022, 04:19:12 AM by KK4GMU »
Logged
IC-7100, RSPdx, AT-D878UVII-Plus HT, TGIF Spot

WA3SKN

  • Member
  • Posts: 8126
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #6 on: July 18, 2022, 09:01:42 AM »

I would disagree with most of you post concerning audio, but audio is both relatively cheap and fun to play with... so go for it!
However, this does not address your real problem of a local noise source in your area.  Are you ready to build a directional antenna and try to find and fix your local noise problem?

-Mike.
Logged

KK4GMU

  • Member
  • Posts: 366
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #7 on: July 18, 2022, 09:21:14 AM »

I would disagree with most of you post concerning audio, but audio is both relatively cheap and fun to play with... so go for it!
However, this does not address your real problem of a local noise source in your area.  Are you ready to build a directional antenna and try to find and fix your local noise problem?

-Mike.
Nope!  Repeat after me:  "HOA", "very small lot", "no trees", "no attic access", "topic of thread is post RF audio", "not interest (right now) in RF "solutions", "current focus: what works best in audio domain."
« Last Edit: July 18, 2022, 09:26:20 AM by KK4GMU »
Logged
IC-7100, RSPdx, AT-D878UVII-Plus HT, TGIF Spot

K7JQ

  • Member
  • Posts: 2602
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #8 on: July 18, 2022, 09:53:12 AM »


Nope!  Repeat after me:  "HOA", "very small lot", "no trees", "no attic access", "topic of thread is post RF audio", "not interest (right now) in RF "solutions", "current focus: what works best in audio domain."

Sounds like my situation, except that I have attic dipoles and fixed-mounted screwdriver antennas outside. All you can do is read comments like mine above about the CLRdsp, and others that may contribute experiences with the other "post RF audio" products you mentioned. Then go with what you feel will best help your noise situation. I tried my best to answer your question, but you'll never know until you try one out.
Logged

KK4GMU

  • Member
  • Posts: 366
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #9 on: July 18, 2022, 11:08:31 AM »

I like the fact that there are physical, readily accessible controls on a stand-alone dedicated DSP.  No need to go 3 or 4 levels into the 7109's menu system to tweak the audio.

I'm considering either the CLRdsp or the Xiegu GBR-1, which is a newer version, but fewer reviews.  They are both about the same price.
Logged
IC-7100, RSPdx, AT-D878UVII-Plus HT, TGIF Spot

K6AER

  • Member
  • Posts: 7159
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #10 on: July 18, 2022, 12:16:44 PM »

Your getting the cart before the horse.

Get your General license.

Borrow an HF radio and put up an stealth antenna at your location to see what your noise floor is on 80-20 meters. You may fine your location is completely unusable for real HF work.

If you buy an  HF radio make sure it has at least a 2 KHz adjacent selectivity of -95 dB. or better. This will help recovered audio more than any digital magic box.

Consider what your RFI will be doing to your neighbors.  Do not announce to the locals you are on the air.

With the basics out of the way you can adequately play with audio noise reduction with much better results.

Good Luck.

Mike K6AER
« Last Edit: July 18, 2022, 12:21:39 PM by K6AER »
Logged

W6MK

  • Posts: 4095
    • HomeURL
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #11 on: July 18, 2022, 12:25:35 PM »

I'd like to stick to post RF solutions. Tried it. In my case no cigar. Thanks.

The point many of us are making is that your fantasy about improving clarity of receive speech solely via audio processing is unlikely to bring much, if any, benefit.

Dealing appropriately with your RF/EMI environment would, of course, be a real learning experience, with doubtless great potential benefit. Solving some kinds of ham problems requires going beyond buying another gadget.
Logged

KK4GMU

  • Member
  • Posts: 366
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #12 on: July 18, 2022, 12:30:41 PM »

I like the fact that there are physical, readily accessible controls on a stand-alone dedicated DSP.  No need to go 3 or 4 levels into the 7100's menu system to tweak the audio.

I'm considering either the CLRdsp or the Xiegu GBR-1, which is a newer version, but fewer reviews.  They are both about the same price.
Logged
IC-7100, RSPdx, AT-D878UVII-Plus HT, TGIF Spot

KK4GMU

  • Member
  • Posts: 366
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #13 on: July 18, 2022, 12:41:02 PM »

To the off topic folks, as well-intentioned as they may be:  The off topic comments about RF are so unwelcome and useless to me.  Save it for the RFI/EMI topic, which this is NOT.  I am not you.  Their comments remind me of the little old lady wanting to cross the street, and the "helpful" young fella takes her across the street she just came from.
« Last Edit: July 18, 2022, 12:53:06 PM by KK4GMU »
Logged
IC-7100, RSPdx, AT-D878UVII-Plus HT, TGIF Spot

N8YX

  • Member
  • Posts: 2449
Re: Audio output DSPs and Parametric EQ speech enhancement
« Reply #14 on: July 18, 2022, 12:42:53 PM »

As a general rule of thumb, it's easier to deal with noise before it makes its way into the receiver's AGC loop. The most effective way is to handle this with an RF-DSP front end; no audio DSP made is going to effectively filter out wideband S9+ noise and allow you to grab a desired S1-S2 signal from underneath.

With analog receivers, I use noise cancellers (MFJ-1026) and an amplified sense antenna mounted at rooftop level or higher. I found a Mckay-Dymek unit and run a 60-100" whip on the external mount. Its controller features an attenuator plus preamp so I can tailor the gain to the day's requirements. The sense antenna is fed by multicoupler to 2 or more -1026s; each of these connects to a "problematic" transceiver as needed.

There are a number of ADSP units in use here. A couple of DSP-599Zx do a decent job at cleaning up ambient background noise. Two of my R-7000s (themselves modified fairly extensively besides) have SGC ADSP-LO units installed internally. These make use of the Speech button to switch between filter settings. I'd give them a bit of an edge at cleaning up a noisy background over the Timewaves, but the latter are much more flexible.

For serious (!) listening, I have a couple premium-grade receivers which are rack mounted. Also in the racks are Tektronix TM-506R power units, which contain a variety of Digital Audio Corporation's lab-grade audio processors. These are somewhat unique in that they can be daisy-chained and one can act as a controller for the rest. One or more per receiver is employed and an SC-502 scope in each of the -506s can look at receiver IF, input to the DSP or output from it. DAC processor units are usable past 7KHz; the receivers are fitted with several filters - independently selectable - and the widest of these is 8KHz B/W.

Yet another set of choices to consider.
Logged
Pages: [1] 2 3   Go Up