ALL ABOUT SPEECH PROCESSING

GIVE YOUR CB RADIO THAT EXTRA “TALK POWER!”

by Lou Franklin
President, CBC INTERNATIONAL

 

Have you ever been watching your favorite TV program, when suddenly the commercials came on and seemed to knock you right off your seat? Or maybe you’ve noticed how most AM or FM broadcast stations playing Rock music always seem louder than other stations? Well, they haven’t turned up the power or program volume! Instead, they’ve applied a form of audio or speech processing. You can also take advantage of such techniques to make your CB radio or HAM radio signal louder and more powerful. In this report we’ll discuss the basic methods, as well as the pros and cons of each.

 

SIMPLIFIED BACKGROUND THEORY

It’s well known that the human voice is quite weak when compared to other types of audio signals. For example a solid tone, which is a sine wave, has about twice the power to modulate a radio transmitter than the average voice. Unlike the familiar sine wave with its regular repeating pattern, a human voice is a very complex waveform having many "peaks" and "valleys." The peaks represent the points of maximum transmitter modulation, and the valleys the minimum.

The problem is that those sounds producing maximum modulation are not the same sounds that give maximum intelligibility of the transmitted signal. ("Intelligibility" means how easily the voice can be understood, especially during conditions of weak signals or heavy interference.) These high-intensity, low-intelligibility sounds are the familiar vowel sounds we know from grade school: A, E, I, O, U. It is the consonant sounds, such as B, K, L, S, T, which actually give the voice its greatest intelligibility. So if we can increase these consonant sounds relative to the vowel sounds, the overall average voice power will be greatly increased. This is illustrated in the ’scope photos of Figure 1.

FIGURE 1
EFFECTS OF SPEECH PROCESSING ON PERCEIVED SIGNAL STRENGTH

Photo: Effects of Speech Processing (62446 bytes)

The greatest amount of voice power of benefit to radio communications is concentrated in a rather narrow band of frequencies, from about 500-2500 Hz, and centered around 1000 Hz. Women’s Lib notwithstanding, the male voice is more powerful because it contains more of the frequencies near the 1000 Hz median, while female voices contain more of the higher frequency elements. If you’ve ever watched a TV show or a movie being filmed, you may have noticed this; in a shot with a man and a woman, the soundman always hangs the mike boom closer to the woman to keep the level of both actors’ voices equal. So a second but equally important part of speech processing is to control the frequency response. We want to filter out all the low-power voice frequencies, and concentrate only in the 500-2500 Hz range. This is also important because it limits the signal bandwidth. In other words, we don’t want to be heard on three channels at once!

MEASURING SPEECH PROCESSING EFFECTS

Since this discussion deals with power levels, it’s not surprising that speech processing and its effects are measured by the familiar "decibel" or "dB." In its simplest terms, a change of "so many dB" means that the signal power has been multiplied (or reduced, for that matter) by so much. Here are a few power multiplication factors:

+3 dB = power multiplied by 2
+6 dB = power multiplied by 4
+10 dB = power multiplied by 10
+20 dB = power multiplied by 100

For example, if your transmitted signal increased by +3 dB, this would be the same as doubling your power. (Multiplying by 2.) The difference at the other guy’s receiver is the same as if your transmitter power increased from 5 watts to 10 watts. But this +3 dB increase was the result of speech processing.

There are two important measurements related to speech processing:

1. The actual amount of processing applied to the signal, expressed in dB.

2. The effective increase in intelligibility of a processed signal over an unprocessed one, also expressed in dB.

Suppose the power of the consonant sounds were increased ten times relative to the vowel sounds. That’s the same as saying that 10 dB of processing was applied to the signal. The amount of processing can be adjusted by the operator with a potentiometer control in most speech processors. This adjustment effects intelligibility and the degree to which the voice changes from its unprocessed natural sound.

The true test is the actual increase in readability over an unprocessed signal, all other factors being equal. To say that processing adds 3 dB of intelligibility is just like the transmitter power doubling described above. The person hearing you can’t tell the difference between a doubling of RF power, or a doubling of effective power due to the speech processing. It’s important to remember this when considering the various processing systems that can be applied to your CB or HAM rig.

 

METHODS OF SPEECH PROCESSING

There are four basic systems that have been used to increase the "talk power" of an AM or SSB signal.

1. RF CLIPPING

2. RF COMPRESSION

3. AUDIO COMPRESSION

4. AUDIO CLIPPING

Let’s briefly examine the pros and cons of each method as it might be used in the typical HF transceiver. The effectiveness of each has been summarized in the graph of Figure 2, which we’ll refer to often.

FIGURE 2
EFFECT OF PROCESSING ON SIGNAL-TO-NOISE IMPROVEMENT AT RECEIVER

Figure 2 (graph) (9066 bytes)

 

1. RF CLIPPING

As the graph clearly shows, this is by far the most effective method. But it’s restricted just to SSB transmitters. That’s because the processing occurs in the IF stage of the transmitter, after the SSB signal has been generated, rather than in the audio stages. A typical system is shown in Figure 3. Diodes are used to "clip" or chop off the high-intensity peaks of the RF waveform itself, resulting in a higher average power relative to peak power.

 

FIGURE 3
PLACEMENT OF RF CLIPPER IN TYPICAL SSB TRANSMITTER

Figure 3 Block Diagram (2451 bytes)

Figure 2 showed that about 20 dB of RF envelope clipping can result in an incredible 8 dB improvement in signal intelligibility. This is equal to a power multiplication of over six times, which is about the same power advantage of SSB over AM in the first place!

We never get something for nothing, including electronics. Clipping of any type produces lots of "harmonics" or undesirable signals. No doubt you’ve already experienced this as "splatter" or channel "bleedover" interference. Just like overmodulation. So any clipper circuit must be followed by some kind of filter to prevent the clipped signal from being heard all over the band.

The disadvantages are cost and installation complexity. While RF Clipping is the most effective, it’s also the most expensive, in the $140-$175 range for typical HAM-type RF Processors. And since the circuit must be installed internally in the IF stage of the transmitter, it takes a skilled technician to do it. Not your simple add-on Black Box! You’ll only find built in RF Clipping in the very top-of-the-line HAM rigs, like the Kenwood TS-940 in my own shack.

 

2. RF COMPRESSION

This type of processing is commonly found in virtually all CB and HAM SSB rigs. You may have heard it called by its more common name, "ALC," or Automatic Level Control. Basically this is a simple feedback system identical in principle to the Automatic Gain Control or AGC circuits used in receivers. A small RF signal sample is picked off after the final amplifier stage, and fed back to an earlier RF stage to control its gain. The stronger the incoming mike signal, the more the RF signal to the antenna is reduced. The result is to keep the output signal level constant, even though the input power of the operator’s voice may be changing a lot. This method is shown in Figure 4.

 

FIGURE 4
TYPICAL RF COMPRESSION (ALC) SYSTEM IN SSB TRANSMITTER

Figure 4 Block Diagram (3110 bytes)

The true purpose of ALC is not really to increase talk power, but to prevent bleedover interference. When the signal to the final RF amplifier stage is too strong, that stage is overdriven and the result is adjacent-channel splatter. (This is the familiar "flat-topping" of signal peaks you see on a ’scope in a misadjusted SSB rig.) As shown earlier, this method only adds about 1 dB of intelligibility to the SSB signal.

 

3. AUDIO COMPRESSION

This is the most common system. It’s found in many popular CB power mikes, and a form of it is used in all 40-channel and most newer 23-channel CB rigs. (Where it’s called "AMC" or Automatic Modulation Control.) This is the same idea as the ALC or RF Compression method, except here the sample signal is fed back from the last audio stage, not the RF final stage. The real purpose once again is not to increase talk power, but to prevent overmodulation on voice peaks and its resultant bleedover interference. See Figure 5 below.

In the case of the compression type power mikes, the voice signal has been made constant in level before it even enters the radio. This is done by amplifying the weaker voice elements in relation to the stronger peaks. One obvious disadvantage here is that such circuits are very sensitive to things like background noise, or AC power line hum in the case of base station rigs. A listener may hear your kids yelling in the next room, or your voice may sound very hollow if you’re not speaking closely into the mike. In any event, the graph of Figure 2 showed that audio or "volume compression" adds almost nothing to signal readability, only about 1 dB.

FIGURE 5
TYPICAL AUDIO COMPRESSION (AMC) SYSTEM IN AM TRANSMITTER

Figure 5 Block Diagram (2793 bytes)

 

An interesting variation of audio compression is the so-called "logarithmic" speech processor. In very simple terms, this is a smart processor which knows which voice elements to amplify and which to ignore. The stronger the incoming voice signal, the less amplification is applied, and vice-versa. You’ll pay a big price for this gadget too, around $75. And like regular compression circuits or power mikes, the net increase in intelligibility is not that impressive. They’re also very sensitive to background and AC power line noises. Sophisticated noise-blanking and filtering circuits are required, which partly account for the high price tag.

It’s important to note here that many so-called "power mikes" have no processing circuitry at all, not even compression! Instead they amplify all voice elements equally. As you should know by now, any system that doesn’t change the audio or RF peak-to-average relationship does nothing to increase your real talk power.

 

4. AUDIO CLIPPING

This is the simplest and dollar-for-dollar most effective method. Figure 6 shows a basic system. The mike signal is first amplified to a level high enough to clip off the high-intensity, low power peaks. One or two diodes are used for this. The range of voice frequencies are filtered to meet the required 500-2500 Hz band limits. The processed signal is then reamplified and the undesirable harmonics are removed by filtering. Finally, the signal is adjusted to a level which will produce 100% modulation of the AM/SSB transmitter.

 

FIGURE 6
TYPICAL AUDIO CLIPPER SYSTEM

Figure 6 Block Diagram (2702 bytes)

Looking again at Figure 2, you see that this method is the next best substitute for RF envelope clipping. But it’s a cheaper, easier to install, and will work with any AM, SSB, or even FM transmitter. (Remember the TV commercials? TV sound is FM.) Using 15 dB of audio clipping, you’ll get about 4 dB of real increased intelligibility. This represents a 2 times power multiplication factor; a 4-watt CB sounds more like 10 watts!

With clipping increased to about 25 dB, you’ll squeeze out another 1.5 dB of readability. At such high clipping levels the voice starts sounding unnatural, but is quite readable during weak-signal or heavy interference conditions.

Since clipping of any type produces harmonic bleedover, filtering is an absolute must. This is easily done by adding a few circuit parts to form a low-pass filter, just like the LP filter often added to your base antenna to help reduce TVI.

Another advantage here is that background noise and hum aren’t nearly as noticeable as they are with audio compression circuits. That’s because this method doesn’t amplify the weak sounds in relation to the strong ones, but rather reduces the strong sounds relative to the weak ones. The difference is very subtle, but the results are often a real increase in talk power.

Perhaps the greatest advantage of audio clipping is its price and simplicity. The circuit is connected in series with the mike audio and permanently wired inside the radio. It can also be added externally in a separate box if desired, but doing this increases the risk of stray RF pickup because of the longer wires needed. If your CB happens to have a MIKE GAIN control, you can conveniently adjust the clipping level using that. The low price makes it cheaper than most power mikes. And as we’ve seen, more effective in producing a genuine power advantage.

Our company, CBC INTERNATIONAL, has offered this useful accessory for many years now. You can order the complete kit ($38), or just the plans & PC board ($12). It’s designed for people who want to get the maximum performance from their rigs. See our separate Web page on the DYNAMIC SPEECH PROCESSOR.

CBC INTERNATIONAL   bullet-1.gif (124 bytes)  P.O. BOX 30655   bullet-1.gif (124 bytes)  TUCSON AZ 85751 U.S.A.
TEL/FAX: 888-I-FIX-CBs (1-888-434-9227), (520) 298-7980
Electronic Mail:
To send us an email now, click
here.

ORDER DSP KIT SECURE NOW!         HOME          CONTACT US