Everybody likes a good voice acting joke.  Part of the charm and legacy of Resident Evil has always been its terrible (and terribly-delivered) lines.  We’ve been laughing about Jill sandwiches and masters of unlocking for two decades.

I used to wonder how those lines made it into production in an otherwise high-end (and generally well-translated) game.  Perhaps there were no native English speakers involved with the VO production? Maybe the team just didn’t care?  In some deep trench in my brain, beyond the reach of my skepticism filter, a little whisper wondered if the poor voice acting in the original Resident Evil might have been intentional.  Could the latent ridiculousness of that entire series perhaps be the product of design and not just a reflection of American media in a warped Japanese mirror?

This entire train of thought jumped the tracks when I started adding voice acting to my own games.  The answer was suddenly very clear: good voice acting in video games is hard!  Like, really hard!  It’s hard to write good dialog, it’s hard to find (and direct) a good actor, it’s hard to make a 3D game character emote in a believable way while playing a sound file that is supposed to be words coming out of their mouth.  If you’re not careful, there could be a massive delay between the writing of the script, the recording of the dialog, and the actual integration of the recording into the game.  By the time you figure out it’s not working very well, it might be too late (or too expensive) to change.

It was with more than a little trepidation that I approached voice acting for DEAD SECRET CIRCLE.  I had fun writing the dialog system itself, but found the prospect of getting lines recorded daunting.  With text, I can noodle on the word selection or the speed at which information is released to the player all the way up until the launch of the game.  But with voiced dialog, I had to commit to my own writing a lot earlier than I wanted to.  My fear that the actual words would sound lame once played in the game was unassuageable.  Nothing destroys the mood in a horror game faster than a poorly-delivered line.

Finally I did the sensible thing and turned to an expert for help.  I called my good friend and fellow narrative game design geek David Chen.  David wrote the story and ran production for Narcosis, and he’s quite the veteran when it comes to dialog and voice production (look for his name in the credits of any recent Metal Gear Solid game), so I asked him to handle recording and editing for DEAD SECRET CIRCLE.

David sliced into my text like a surgeon removing a tumor.  Editing for vocal performance seems to be a completely different skill from editing for readability. He inserted strategic ellipses, added notes about emphasis and context, and removed weird words that nobody actually says out loud to convert my ~300 lines of dialog into an actual script.  We recorded the entire game over the course of two days, using actors and a (fantastic) engineering team David sourced.  This was the point at which I had to finally let go of my dialog and let it get chiseled into stone.  I was thankful to have an expert to run the session for me.

Something weird happened over the course of the recording: my dialog took on a life of its own.  I have read these lines to myself in my head a thousand times, but coming out of the actor’s mouths they suddenly sounded different.  The performances were stellar, but it was more than that.  These weren’t my words anymore–I was hearing the voice of my characters.  The sensation was very strange, a sort of deja vu, like seeing your reflection wink.  I guess this probably happens to script writers all the time, but it was a first for me.

After recording wrapped the lines were cut up into files, integrated into the game, and hooked up to animation, and we saw our completed characters for the first time just a few weeks before the game was scheduled to be released.  The result was shockingly better than I had anticipated, even though we’d had placeholder VO integrated for months.  Now that the game is out and we can see the reaction of real players, I feel even better about how the dialog, VO, and character animation came together.

But it could have gone the other way.  Dialog recording always comes late in the project development cycle because there are just too many other parts of a game that need to be operational before the final script can be decided upon.    Dialog recording is expensive and time-consuming to redo, and yet the overall quality of the finished scene can’t be properly assessed until all of the lines are in place.  I wonder if bad voice acting in otherwise high-end games is the result of time pressure more than anything else.  Maybe by the time those games realized that something was wrong it was simply too late to change.