It was written by John Catalano, PhD. in his regular review section "Computers & Radio".
I have the text here but the "Figures" did not scan well and I have left them out.
I would like to thank Dr. Catalano and Monitoring Times Magazine for their interest in my program.
Today, although the shortwave bands are not as crowded, heterodyne whistles and electrical noise can still take all the fun out of any monitoring. This month, we are going to look at two computer programs which may herald the end of whistles and noise: the bane of monitors everywhere!
Audio filtering comes in many forms: passive components, active elements, and most recently, digital filters. The first of these filter methods goes way back to the beginnings of radio. It uses time constants associated with the interaction of inductors, capacitors, resistors and electrical signals. Using specific fixed component values chosen to eliminate a specific interfering frequency, or range of frequencies, filter circuits could be built. Banks of filters were employed where frequency flexibility was required.
The second method became popular in the nineteen seventies as integrated circuits allowed easy and economical use of high quality operational amplifiers. These amps are the building blocks of active filters. They enhance the behavior of the passive components resulting in active filters. A later variation, which led the way to future filtering methods, used active circuits to "act" like passive components. The unique feature of these filters, such as switched capacitor filters, was that they were "tunable." No longer were filters fixed to given frequencies or characteristics.
As semiconductor engineering became more advanced, the simple integrated circuit evolved into the microprocessor. As microprocessors' capabilities increased in speed and complexity, near real-time data processing became a reality. The digital signal processing (DSP) world of compact disc, laser disc and DVD was born. This technology was applied communications in the form of DSP chips. These devices use software routines, or algorithms, to manipulate and filter signals.
An audio DSP chip is connected between a receiver and a speaker. First it converts the audio into a digital data. Once in the digital domain DSP chips can simulate high quality audio filters of most any type via software routines. Pretty nifty. Then the "filtered" digital signal is reconverted back into audio -sans whistle or noise.
This time we'll look at two programs for your computer which give you DSP audio filtering functions for your scanning, medium or shortwave monitoring. If you have a 200 MHz Pentium computer with a sound card you already have all the hardware.
The first DSP software audio package is the Swezey Digital Filter Program We used the standard (non-stereo) licensed version 220.127.116.11. All you will need is MS-Windows 95/98/NT running on a 200 MHz (or faster) computer, a sound card and a receiver. As usual, when using a receiver and a computer, it is recommended to employ a shielded antenna in order to minimize computer-generated noise. No installation is required other than copying the program to the desired directory. Now all we have to do is to connect the external speaker/Line Out of our receiver to the Line In of our computer's sound card, and we are ready!
First - Be careful to minimize all volume/level controls: the receivers, the computers and the external power amplifier (if you use one as I do). If you don't you could blow out your speakers. Then turn on your computer and your receiver. Tune to a broadcast station for use during setup. Increase the input level of your Line In until you hear the broadcast station. Usually this is done via right clicking the speaker icon on the lower right of your screen.
Now let's run the Swezey DSP Program. The entire program is controlled from one screen, which displays the audio spectrum of the Line In signal. A screen similar to Figure One will be displayed. Here, we are tuned to a frequency in the medium wave band where two adjacent stations are making life noisily difficult. The culprits are visible in Figure One as two "spikes"; the large one at 1,000 Hz and a smaller one at about 2750 Hz. Oh yes, Deja vu! The painful whistles bring me back twenty years to the FRG-7!
The Swezey Digital Filter can attack this cacophony in a number of different ways. The easiest, totally automatic method is via the Options command menu, right above the frequency markers. See Figure One. Clicking Options brings us to the Auto Heterodyne Reduction (AHR) menu. Then choose Fast (Voice), since we are attempting to listen to talk programs. See Figure Two results. Notice that the two spikes, seen in Figure One, are covered by broad vertical bars. The Auto Heterodyne Rejection function has found the frequencies having constant amplitudes which correspond to our whistle. Then it has applied notch (band stop) filters to them, indicated by the broad vertical bars. The result is beautiful, whistle-less audio. No headache here!
If you look carefully, you will see other broad bars. The AHR has found smaller spikes and automatically notched them out as well.
Alternatively, we could have manually notched out the offending frequency ranges. We can generate notches by positioning the cursor to the left of a spike. Click and hold the left mouse button. Now move to the right of the spike and releasing the left mouse button. A broad vertical bar appears over the spike. Right clicking on this area, allows us to make it a Stop Band (notch). See the middle region of Figure Three. With the Swezey program you can manually setup a number of these notch Bands. Figure Three shows this manual method on a 2850 Hz "whistle." And yes, this signal then became listenable.
Under the FFT Filters menu (Fast Fourier Transform - a mathematical methodology) there are thirteen predefined filters. Three are specifically for voice applications and two are for music. A notch, three high pass and four low pass filters, each at a fixed frequency, can also be employed with just a mouse click.
It does a great job making an interference-ridden signal listenable. The Swezey program is very easy to use and worked without a major problem. Some small glitches were noted. The display sometimes seems to do strange things for a few seconds. It's not a real problem since the filter's operation is not affected.
The automatic level function, sometimes, doesn't quite get it right. The effect is very low audio output. However, I found that this is easily remedied by going to the manual level control setting, and then back to automatic.
Swezey also provides other filter types and features that we haven't covered. The flexibility and usefulness that the Swezey program provides far out-weighs these operational quirks.
Earlier, I stated that DSPs perform "near real-time data processing." Well, the "near real-time," in the case of the Swezey program, adds a delay of over 1.5 seconds on a Pentium 1,233 MHz computer. This delay makes tuning while listening to the Swezey filter's output impossible. But all is not lost.
All it takes is the ability to monitor the receiver's audio before filtering. This can be accomplished by selecting the Unfiltered Input on the Audio levels Control menu. See Figure Two. Once a signal is tuned in, choosing Filtered Output will allow you to use the filter.
GNASP1 is an experimental version of a DSP based audio signal processor. The program requires a 200 MHz PC running Microsoft Windows 95, Windows 98 or Windows NT. As stated in the instructions, "It is absolutely essential that a full duplex driver is used" with your sound card. These days, duplex sound card capability is the norm. GNASP1's command structure is far less elegant than Swezey's, relying on a single list of fixed filters. See Figure 4. Sixteen filters are predefined: eight low pass, six pass band and two high pass.
Once the filter is chosen the program displays two screens. See Figure 5. The top screen shows the audio spectrum prior to filtering. The bottom screen indicates the audio spectrum after going through the filter. In this case, we have applied a band pass filter, centered at 800 Hz and having a 100 Hz width. All other frequencies will be highly attenuated, as the bottom screen indicates with a Peak around 800 Hz.
GNASP1's installation was also very easy. Each of its sixteen filters worked well. The additional screen showing the resulting spectrum is a nice touch.
GNASP1 also has a user configurable filter capability, and comes with 2200 Hz low pass filter. The instructions indicate that with the "appropriate program" the user can program custom filters. But it does not give information on where the "appropriate" program can be obtained.
As with Swezey, GNASP1 delayed a real-time input by about 1.5 seconds. This again can be over-come by first selecting "Direct mapping of input to output" on the filter list. Once a station is tuned in' then you can choose a filter from the menu.
Although GNASP1 worked well, it is clearly a work in progress, as it states in its instructions. But since there is no charge for GNASP1 you can use it to see if DSP filtering is for you. If it is, it may be enough. If not, then the $50 for Swezey Digital Filter will be worth it.
Download GNASP1 at http://members.tripod.com/~gniephaus/gnasp1/gnasp1.html.
The Swezey website at http://www.winternet.com/~swezeyt/dsp.html will allow you to download a "crippled" version which has sporadic noise bursts. You can also get ordering details for the full, uncrippled, version at this site.
Well, there you have it. In the future we'll look at other audio DSP programs which are available. Give audio digital filtering a try in your monitoring. It's better than aspirin, or a divorce lawyer... Come to think of it, anything is better than the latter!
(c) 2000-2009, Thomas F. Swezey. All rights reserved.