This year I was fortunate enough to go GDC and attend the audio track. I took copious amounts of notes in point form, so I figured I might as well share them here. Some of them may not be entirely comprehensible, hopefully I can figure them out as I go along. Points marked with a "•" are basically what my notes where. Anything with a "-" is me trying to explain my note post-hoc.
The first talk I attended was Sound Design the Bungie Way, by Jay Weinland.
• When you have too many things to assign unique sounds to, try to solve it with UI
- Jay was talking gun sounds here, I think the idea here was that the acutal firing sounds for the guns are so design intensive, giving them different reload and ammo PU sounds can help give them a unique sonic identity
• Think of sounds as being attached to the player, or detached from the player. The former should be 2D sounds, the latter, 3D sounds
• Bungie removed weapon reverb tail from shots themselves, and assigned them to envinronment type and surface type. Tails were broken into three basic weapon type (shotguns, assualt rifles and sniper rifles) then broken down by environment, then by surface type
- reverb talks seemed to use the same source sounds, but were treated with EQ (and additional in game reverb) when they were further away from the player
• Apparently you don't have to get special permits (beyond gun permits) to use tannerite to create explosions (for recording)
• Bungie augmented all their gun recordings with a transient layer off the top and a "punch", "snap" and "lowend beef/mech" layer as well
• Enemy guns had two basic variation sets, distant and close. Because the vast majority of gun play happens at the middle distnace they made the decision not to change between distant and close sound at the mid point, to much switching between sounds all the time. Instead they crossfaded into distant sounds further away. For the mid range, they used the close sounds with some HPF and LPF.
• a mix of three global dynamic systems were used for state mixing
* Ducking - gentle slow release, a few db ducking happens of incoming input level
- presumably this was to prioritize medium/high priority sounds without a lot of volume
•Sidechan - working at about 6db of gain reducion,
- input dependant, so this makes more room for loud sounds
• Silence Duckers - aggressively kills everything for an instant to leave room for just a huge explosion
- So the way I understand the silence plugin to work is that WWISE just looks at the level it is set at as opposed to analyzing the sound it is at. By making it super short, setting its level to the max and having it play just before your explosions in an explosion event, you can kill all other sounds faster that you can hear them cut out giving room for the explosion to really cut through.
• Instead of a long continous loop, rocket by's were treated as long one shot sound that played in proximal relationship to the player. If the rockets were fired very close to the player, the audio engine would seek into the middle of the pass by audio file, quickly fading up.
- I followed this conceptually, but Jay didn't have time to get into how it was scripted.
• Consider using state mixes (ie scopes affecting pan)
- I litteraly just remembered that "scopes affecting pan" refers to using a separate mixing state for each gun scope and narrowing the pan to help the feeling of focus for looking through the scope.
Okay, with any luck, I'll be able to summarize another talk for y'all tomorrow.