Sound System Design Considerations
The sound from a good Pro Audio Sound System should be:
• Evenly dispersed
• Adequate volume
Clean sound is free from distortion. Although there are other forms of distortion, the most recognisable kind is characterised by the "fuzziness" and/or "roughness" that is created when amplifier circuits are overloaded. Although this is desirable in some situations (an "overdriven" electric guitar sound, for example), it is very undesirable in a live sound reinforcement system: our aim is to reproduce the "overdriven" electric guitar sound accurately, not to add distortion that wasn't already there.
Clear sound is free from electrically generated noise. If the PA with all the inputs muted sounds like Niagara Falls (or like the engine room of a supertanker), the performance - and the audience! - will suffer.
Sound is intelligible when all its components are clearly recognisable and comprehensible. If someone is speaking clearly into a microphone, to anyone in the audience they should sound like someone speaking clearly (they should be intelligible). Where poor room acoustics limit the range of intelligibility (see critical distance), a good system properly set up should nevertheless succeed in producing sound that is intelligible for most of the audience.
If some frequencies are predominant or lacking, the resulting sound will be unnatural. For some special effects this may be desirable, but in general we don't want the singer to sound as if s/he is using a cheap megaphone, or the kick drum to sound as if someone is banging on a wooden door with a rubber hammer. If the singer is actually using a cheap megaphone, however, we want that to be reasonably obvious to anyone who has their eyes shut. Although equalisation can compensate for some instrument qualities, a budget plywood third-world guitar naturally sounds like a budget plywood third-world guitar, and there are limits to what even the best sound system in the world can do with it, especially if the pickup is a budget third-world transducer. Similarly, very expensive microphones, mixer, amplifiers and speakers will make a badly-tuned undamped drum-kit sound exactly like a badly-tuned undamped drum-kit.
A balanced sound is one in which all the separate elements contribute. If the guitar is so loud you can't hear the rest of the band clearly, or if the snare drum is obscuring the vocals, the audience probably won't get the best night of their lives. Balance is important for the audience, but can be less useful for individual musicians, who need to concentrate on their own or other specific instruments (which is why even small systems often need monitors). Guitarists (even guitarists in the audience) often think the guitar isn't loud enough. The same applies to bassists and basses, fiddle players and fiddles, and you-name-it players with you-name-its. Although there is a great deal of personal taste and subjectivity involved, the aim is balance, not even more of the musician who makes the most noise or is going out with your sister.
Evenly dispersed sound
We don't just want clean, clear, intelligible, natural, balanced sound in the place where the sound engineer is sitting: we want everyone to get it. We want even coverage in as much of the audience area as possible. Radical changes in tone as you move through the audience area are undesirable. Our loudspeakers are selected for even dispersion (the sound at all frequencies is reasonably constant within the area the speakers are designed to cover), and good pattern control (sound must not spread to areas where it isn't wanted: sound reflecting from walls and ceilings interferes with evenness, causing peaks and troughs - AKA comb filtering - at different frequencies). Speaker position is an important factor here.
Adequate sound level
Although very high volume can seem impressive, loudness can interfere with all the desired characteristics of a good Sound system (and, indeed, sound that isn't clean, clear, intelligible, natural, balanced, or evenly dispersed - or even, for that matter, adequate in level - can often be described as "loud"). There may also be legal implications if the sound is too loud. It does, however, need to be loud enough.
As well as being loud enough a system needs headroom (we want at least a few decibels spare, to stay well clear of distortion). For that reason, it is better to have more power than you need. The downside of this is the cost involved: all else being equal, you need to double the system size for every 3dB of extra output. If you want 3dB of headroom, you need a system that is twice as powerful as one that has no headroom at all.
The level at which a sound is adequate will depend on the main purpose of the system.
For speech, system requirements are relatively modest (a system capable of producing a peak level of around 85dBA at the listener's position is generally adequate). Headroom requirements are also modest, and output below 100 Hz - or even much below 200 Hz - is unnecessary (and can even be a hindrance).
For recorded music (e.g. typical disco/dance music), higher output and extended bass range are needed. However, there is no need for lots of extra headroom, as the peak level of recorded music is predictable, and the dynamic range of typical disco/dance music is limited.
For live music applications, system requirements are much more demanding. A snare drum can produce levels above 120dB SPL at the microphone. Other acoustic instruments (particularly brass) produce impressive amounts of sound without a sound system in sight. Amplified instruments (especially where the sound is compressed by the amplifier, producing very high average levels) are even more of a challenge. A loud rock band will need a system that is capable both of exceeding the level of the backline, and of delivering it to the audience at much higher than safe listening levels (typically peaking at around 110-115dB SPL at the ears of the audience). Also, dynamic range is typically much greater, which places a corresponding demand on the system's headroom - a requirement of 6dB or more is not wildly exaggerated - even where compressors and/or limiters are used.
Our own Sound systems are generally adequate for audiences in the size ranges stated on each system's page. However, very "loud" styles (e.g. heavy metal) will generally require much higher output capability.
What the sound system cannot do
• Clean up distortion in source instruments or backline amplifiers
If it was distorted before it reached the multicore it will still be distorted when it gets to the sound system.
• Silence a noisy effects pedal or stop your guitar from buzzing near the dimmers
Single-coil pickups can pick up almost anything (including dimmers, induction loops, and the local taxi service). There isn't a "subtraction" button on the desk, and although notch-filters and high-pass switches can reduce the effect of instrument-borne 50Hz mains hum, they can't remove it.
• Make a mumbling talker or incoherent singer intelligible
Poor enunciation and bad microphone technique can be fixed: take elocution lessons, and learn how to use a microphone.
• Make a cheap piezo pickup sound natural
A good EQ section can do quite a lot with the most prominent colouration, but don't expect miracles. Similarly, using your own microphone can be a good idea, but the better the PA system, the more you will notice the difference between your $20 bargain and the other singer's Shure KSM9. Both will sound their best if you have the best PA possible, but there is a lot of distance between them.
• Make a balanced sound out of a ragged unrehearsed noise
Again, the mixer offers some control, but can't completely compensate for poorly-managed dynamics, and certainly can't stop the harmonica player from soloing over the lyrics. A watchful sound engineer can limit the damage, but can't make the overdriven lead solo happen at the right time.
• Disperse sound evenly round corners
Where there isn't a sight-line from audience to loudspeaker (behind pillars, alcoves, partitions, balconies), or where the audience is dispersed over a wider angle than the speaker system is designed to cover, some unevenness (or worse) is inevitable. Covering the whole target area may mean you need extra loudspeakers, as well as additional processing so that time-alignment between separate speakers or speaker clusters is as close as possible.