Feedback is the bane of any audio system, no one likes that high-pitched screaming sound, and yet it’s surprising how often it occurs. Here are some basic, general strategies that can help alleviate the problem in many situations.
One of the most powerful tools in fighting feedback is equalisation or EQ. Equalisation is defined as selectively boosting or cutting bands of frequencies to improve the performance of a sound reinforcement system. Performance is a bit of a vague term, but in this case we’ll take it to mean a combination of intelligibility and overall output level before feedback.
Before getting too carried away it’s important to understand what you can and can’t do with EQ. Firstly, it can improve the naturalness and intelligibility of a sound reinforcement system by emphasising the required frequency ranges for the kind of audio it’s going to be dealing with. Secondly EQ can increase the overall output levels before feedback occurs by reducing the system’s output in frequency bands in which feedback is occurring.
There are several things that you cannot do with EQ. First and foremost to remember is that it will not enable you to make a poorly designed sound system work satisfactorily. Equally it will not resolve intelligibility issues arising from reverberation, reflections, mechanical vibration, high background noise levels, or other problems caused by location or physical design.
EQ also cannot solve problems relating to poor micing technique such as the talker being too far from the microphone, nor can it compensate for poor components in the system. Finally, it cannot improve on echo return problems in teleconferencing systems.
The approach to equalisation as a rule should be softly, softly catchy monkey. After each adjustment listen to the system and assess the results. You are aiming to improve the sound quality as well as improve the gain before feedback. When the system is loud enough and / or clear enough, STOP! Also, if equalising leads to degradation in sound quality, stop and examine the whole system in detail.
The diagrams show the ideal frequency response curves for speech and music. As you can see, the speech curve, isn’t really. It’s pretty flat. Above 2kHz, the response drops off by around 3dB for each octave. Between 125Hz and 2kHz the curve is flat with a window of ± 2dB and below 125Hz the frequencies are rolled off to minimise rumble or boominess. This curve is based on research into human audio perception.
The music curve is much flatter since the range of frequencies experienced is much greater than simply those found in the human voice.
Other solutions:
Whilst EQ can be a powerful tool in the fight against feedback, as already evidenced it has limitations. Feedback is defined as occurring when the amplified sound from any loudspeaker re-enters the sound system through an open microphone and is amplified over and over. In trying to combat it there are several steps that can be taken during system design before you even get to the stage of equalising.
The important term to remember is gain before feedback. This is the level at which you can operate the sound reinforcement system before you run into feedback difficulties. The goal here is to increase the gain before feedback to a level such that the audience can hear and understand the sound clearly without feedback happening.
There are four main things that contribute to making feedback problems worse: Placing loudspeakers too close to microphones, too many open microphones, boosting tone controls indiscriminately and rooms surfaces that are hard and therefore reflective. These include glass, marble or wood.
So, what to do if feedback occurs before the sound system is loud enough? At the most fundamental level you can request that the presenter speaks louder into the microphone, the louder they talk the less you need to amplify the sound and the less the propensity for the system to feed back. Clearly this solution has its limitations, the whole point of a sound reinforcement system is to assist a speaker by amplifying their voice. Shouting is not an option. Secondly, reducing the distance from sound source to microphone works. Halving the distance will increase the system output by 6dB. This can be achieved by instructing the person speaking to position themselves correctly or, in the case of a suspended microphone lowering it closer to them.
Reducing the number of open microphones is another solution. Theoretically, halving the number will allow you to increase output by 3dB. In practice this could mean switching off unused microphones or in the case of a discussion with multiple participants consider shared microphones such as boundary layer solutions. Failing this, look at the type of microphones you are using. Reduce the pick-up field to be more directional by switching from a cardioid to hypercardioid microphone. In an extreme situation, perhaps with a lot of background noise, you could use a differential. This requires the user to brush the screen with their lips to get any response, and excludes any sources that are more than a few feet away.
You could also consider giving people lapel microphones, which will again reduce the chances of sound from stage monitors or the like being picked up.
Loudspeakers also have their part to play. Doubling the distance between loudspeakers and microphones allows output to be increased by 6 dB. Similarly, moving loudspeakers closer to the audience increases the output at the listener by 6 dB. Great, but not always practical. A more satisfactory solution is to use directional loudspeakers. If you are directing the energy away from microphones and towards the audience, then there will be less potential for feedback. Directed sound will also have the same effect as moving loudspeakers closer by increasing the SPL at the listener.
It is of course possible to employ a feedback reducer to kill off the frequencies that are feeding back, which will typically be allow you to increase system output by between 3 and 9 dB. However, you shouldn’t try to rely solely on equalisation or a feedback reducer to provide sufficient additional output in a sound system where the set up itself is poor i.e. microphones too close to loudspeakers etc.
Down to experience:
Ultimately these pieces of advice are no substitute for experience. There are a huge number of tricks that audio-professionals pick up over the years for particular situations, venues or pieces of equipment. However, the principles laid out above apply in almost every situation and will go a long way towards solving problems.
A final note of caution regarding electro-acoustic modelling software. It’s very easy using modern packages to take your venue and design a beautifully smooth system with near perfectly uniform coverage. However, is that really what you want, and does the system design pass the common sense test? It is after all possible to produce such coverage by taking a single loudspeaker and pointing it at the ceiling!