RSL - Audio Processing

Overview Products by manufacturer Broadcasting Streaming Telemetry RSL / Hire

Audio Processing

Why do I need audio processing?

The main reason that audio needs to be processed prior to transmission is that the transmitter can only accept a limited input signal range up to a maximum of 100% peak modulation. Your RSL licence strictly forbids 'over-modulation' because this causes interference to users on adjacent channels, and causes your sound to become distorted. On the other hand, it is no good just turning down the input to the transmitter, because with high modulation the signal will go further and the better it will sound on the fringes of its coverage area.

Programme controllers and station managers want their station to sound loud compared to the next station on the dial. An engineer might say 'what is the volume control on a radio for?' but programmers see a competitive world out there.

Most radio listening takes place in the car and on portable radios. To get the best out of the cheaper models, the bass and treble in the signal need to be emphasized, but in a way that doesn't make the station ever sound muffled, tinny or distorted over a wide range of music types.

Finally, a good audio processor will cope with the whole range of levels coming from the mixing desk. These will be subject to wide variation for many reasons, not least that many self-op DJs don't have the time (or, maybe, the inclination!) to regulate the level on the desk as well as they should.

How does a typical audio processor work?

There are a vast number of possible ways of controlling the level of an audio signal. Let's look at a few of the simple ones, and as their limitations become apparent then work on to some of the more sophisticated techniques.

The simplest way of controlling level is to simply chop the peaks of the signal off with a device called a clipper. This is the method used in walkie-talkies, and whilst it does impose a very well defined 100% ceiling on the signal level, its brute force nature can also add a lot of distortion to the signal. Clipping is also what happens when you turn the volume on a radio up beyond what its amplifier or speaker can handle and it sounds awful.

Basic method number two is the peak limiter. This is an amplifier that reduces its gain when the input level goes past a certain limit. This ensures that the output never exceeds a given level. The effect is rather like someone very quickly turning the volume down and back up again to reduce the level of a peak such as a drum beat. The amount of gain adjustment and the timing of it are important in limiting, as if they are not exactly right you can hear it in operation, an effect called pumping.

We already have the components for a basic audio processor in place, and many modern transmitters which include a 'built in limiter' actually have a limiter followed by a clipper. Having the limiter first means the clipper can be accurately adjusted for the best compromise between loudness and distortion. Limiters work fine as long as the incoming signal is in approximately the right range, but what if the DJ makes a mistake with the levels?

This brings us on to Automatic Gain Control or AGC. The purpose of AGC is to correct long term and accidental errors in the level of the signal before any other processing is carried out. Like a limiter, AGC circuits continuously monitor the incoming signal level and adjust the gain to keep it in the right range. But instead of reacting quickly like a limiter and 'squeezing' the peaks of the signal, AGC reacts slowly to maintain the average level, allowing brief peaks through. AGC is like having a careful engineer always checking the meters on the mixer.

Continuing with the previous example of a limiter / clipper combination, adding an AGC stage up front allows for best use of the limiter so it is always reducing the peaks a bit for a strong, punchy sound but rarely having to work so hard that it starts pumping.

The next category of circuit is the compressor. A compressor is similar to a limiter, but with a more gentle characteristic which allows the output level to increase a little bit as its input goes up. Compressors are used to give a very consistent sound and to bring up low level detail. They are often fitted between the AGC and the Limiter stages.

All of the devices discussed so far act on all frequencies equally. This gives rise to the effect called 'ducking', where the peak caused by an instrument at one frequency, reduces the level of another instrument at a very different frequency. For example, a bass drum beat can 'duck' the high pitched sound of a piano or keyboard. The solution to this is to split the frequency range up into a number of bands. 'Multi band processing', as it is known, can be applied to any of the devices from the AGC right through to the clipper. Usually it is the compressor stages that are made multi-band, to yield the 'dense' sound required for contemporary formats.

Once one or more of the stages in an audio processor are multi-band, we then have the possibility of adjusting either the input or the output of the various bands. Then we can emphasize the bass and / or treble frequencies as required.

A major development of recent years has been 'overshoot compensation' which tackles one of the problems of the clipper stage. Although clippers produce an absolute 100 % ceiling, subsequent stages in the stereo encoder tend to re-introduce the peaks that were clipped off. This is due to an effect called 'over-shoot' in audio filters and it even affects digital systems! The solution is called 'over-shoot compensation', which cancels the tendency of the peaks to re-appear by pre-distorting the clipped signal. This can let you have about 3dB of extra modulation, which is equivalent to doubling the power of the modulation.

Finally, digital synthesis of the various waveforms used in the stereo encoding process leads to excellent and consistent stereo separation, which gives a very well defined and stable stereo image.