A game is by definition an audio/visual experience. It is usually the visual side of things that get the most attention and so little time is spent on the other half of the equation. While visuals are critical for having a successful game, having great audio is very important. To think this deeper, try imagining Star Wars without John Williams' sound track or James Earl Jones' booming Darth Vader voice. Poor audio on top of great video can really deteriorate production value. The trouble with audio is that it generally comes late in the development cycle and all to often as an afterthought. Another difficulty is that while poor visuals tend to be obvious, poor audio tends to be a silent failure. Star Wars with slightly worse music would create a sense that "something is missing" or "something isn't right" whereas a bad special effect is much more likely to be spotted for what it is.
So how do we approach audio in a space game like Jumpgate Evolution? The most important thing is to have a kick a** audio department. We have a full in-house audio production team along with good external talent who knows how to “articulate” music. As the visuals and tone of the game begin shaping up we make sure the audio department has access to as much information as possible: concept art, story documents and, of course, a playable version of the experience. We then spend a lot of time looking at other forms of media: movies, tv shows, commercials, other games... anything that might inspire. For a space game like Jumpgate Evolution, it can be quite challenging because there are limited references we can take from. Such as how should alien ships sound when they fly past or what should we use for the environment ambient sounds because in reality, space has no sound.
But our goal is to try to get that same emotional response from the game. Therefore a lot of this is breaking down the various parts that make a given scene evoke an emotional response. By removing elements from a scene you can quickly realize what the core elements of the scene are. For example, does a combat scene in Battlestar Galactica still hold up if the music is different, or if there are no special FX sounds, or if you remove the dialog with some radio distortion. Many things are also very subtle. For example we noticed that in many space scenes there are two levels of spoken audio. The primary is audio directed at the viewer (or the protagonist) and then the secondary audio which is voice chatter between external groups.
The next step is to do some basic concept work. This is done by taking a few minutes of unedited game play video and then letting the sound and music guys do whatever they want to make it sound like the experience we are trying to achieve. This is then evaluated by people watching the video and giving feedback. Once we have a few minutes of game play with the audio the way we want it, we meet with engineers and evaluate the feasibility of the various technical pieces involved. By taking this approach we can avoid spending time on features that will not provide the game experience that people find compelling. We also experiment with existing sound and music to see if it's a quality issue. Sometimes the features and tech are fine, but the audio itself simply requires more iteration.
The final thing to consider is when to stop. As with everything in games, it's difficult to know when you are "there," so we spend a lot of time just testing the game. While we want to create a compelling experience at every level we also don't want to keep working and iterating on things that are starting to have diminishing returns. This point can be incredibly hard to evaluate, especially with something as elusive as audio, but there is a sense that you know it when you hear it.
By having an early prototype that has been tested against a sample audience we can be confident that we are aiming towards a solution that will create the experience players are looking for. By comparing the experience to existing experiences in other popular media, we can be sure that we are measuring against the required quality mark. As with everything else in development it's all about prototyping, rapid iteration and testing.