To successfully marry sounds with pictures, we first need to learn how to listen properly. As children we're taught how to read and to write, but not really how to listen; it happens naturally. When we watch TV, listen to the radio, or engage in conversation, our attention is focused and we'll listen. But how often would you stop and listen to what's around you in the same way as you would look at it?
If you stop and listen to your surroundings, you can work out how the sound is constructed, what layers create the overall soundscape, what the perspectives and subtleties of each layer are. Because it's not just one sound, but a combination. Try it.
Is there a conversation - how many people talking? Is the phone ringing, what's the ring tone, how many times has it rung? Has somebody just walked past, what's the floor surface and the soles on their shoes? To create convincing sound design, these layers need to be unravelled, rather than thought of as flat sounds such as an "office" or "train".
There are two key categories when creating sound for pictures - real and unreal. Real is as you would expect to hear were you there with the camera, like the majority of TV programmes and commercials. Unreal is an exaggeration used to add impact, or a made-up sound, because the visual is of something that doesn't exist. The obvious example here is animation.
What is real? It's not generally appreciated that apart from live action and talk shows, almost all TV and film sound is unreal; we may think it's real, but it's actually not. Location sound, if recorded, will generally be restricted to the dialogue. The rest of the sound will be rebuilt from scratch by sound designers and Foley artists. This is done to maintain clarity of dialogue, and to retain control over what is heard. Otherwise, everything would sound just like a home movie. The art of sound design is to recreate the ambient sound to match the visuals, without the viewer noticing.
Marrying "real" sounds to pictures is, therefore, subject to a certain amount of artistic interpretation. A sound designer will not look at pictures and think: "What sound do I need?" They'll think: "What sounds go together to create the soundscape?"
It's as much about what's going on outside the image (the bit we can't see) as the obvious bits that we can. If the visual is a person walking along a deserted beach, we don't just look for a beach sound effect. Within the picture, we look at what kind of beach it is, pebble or sand, what type of footwear have they got on, what they are wearing (it could rustle), is the person foreground or background? Then we ask what can't we see, but need to hear: birds, trees blowing. Should the waves lap gently to the shore, or be large rollers crashing in? With sound we can create impressions, expanding the image for a feeling of more space, to be somewhere busier than we actually are; somewhere bustling underground, or in fact, anywhere.
A simple way of spotting recreated audio is with synchronicity between sound and pictures. I don't mean bad lip sync, but with unnatural timing. Sound travels slower than light, so when something happens at a distance, we see it before we hear it, the two can be several seconds apart. Think about athletics when the starting pistol is fired, from a distance you see the puff of smoke before you hear the bang, which is why you start the stopwatch when you see the smoke, not hear the gun. In the sound dub, sounds will generally be laid in sync with the action, which, although unnatural in real life, is what we've become used to when watching TV.
The "unreal" side of sound design, like animation, is what really gets an engineer fired up; it's like painting a picture with sound. There are no rules, no right or wrong, and it's more fun. Clearly the audio style needs to match the visuals, unless the brief specifies a clash. But you generally have a clear canvas. Once style is ascertained, it's a case of searching for the appropriate sounds and experimenting.
Sounds are sourced from a variety of places, and the newer generation of audio editing systems, such as Pyramix, make experimenting incredibly convenient. The search process can throw up good sounds, but not necessarily in the right order. I'll consistently jump around different sections of the visuals, laying in different sounds as they turn up. Things may seem disorganised, but I promise you they're not!
Recently, I had a great cartoon where a cactus had been sliced in two and was left dangling by a thread. It's accompanying sound was a comedic, rusty squeaking bucket. The client thought this sound wasn't real enough as a cactus wouldn't squeak - we had just seen two doors collapse, a character run through a wall, and two tons of fruit land on a TV. As I mentioned earlier, there's no right or wrong with sound design, just what feels right.
All jobs have time and budget restrictions, so my advice is to get the pictures to your sound designer as soon as possible before your booking. Experimentation is far easier, and less embarrassing, without a roomful of people telling you when something plainly doesn't sound right.
Music always marries well with pictures. Whether it's well known, specially composed or library, music is effective as either as a main feature or as "audio glue" holding the elements together. It can quickly create mood, drama, give a sense of style, age and feeling. It can also bring to life something otherwise dull. It's a simple way of linking scenes and if the visual isn't flowing well, a continuous piece of music can smooth the transitions.
However, it's not unusual after having spent time designing sound for a commercial, that music is added, and gradually each of the effects are taken out until all we're left with is maybe a tenth of what you started with. It can be a long and painful process, but going through it does often prove the maxim "less is more", and in this situation you have to swallow your pride and try to think, oh well, at least we're still getting paid.