Doing gamedev backwards: Devlog #2

The established wisdom for voice over work in computer games is that it's generally one of the very last things you do.

You are supposed to have fully written, deployed, and extensively play tested the game (especially with adventure type games) before you do any of the audio recording from voice talent. The reason is simple - you can't really change audio recordings after they are made, like you can with written dialogue. So if you want to modify a line, add extra lines, or decide to replace any of the voices, you'll have to go back and do the recording again.

These days, that's not such a big deal - most professional voice talent will have their own fancy recording setup in their home, and work remotely. So you can just call them up on Zoom, get them to do a local high-quality recording, and record just one or two lines or the whole lot in one go. Simples!

But I've decided to do this the 'wrong' way.

It bugs me in games (and even TV shows) where everyone is recorded separately, with different mics and acoustics (no room is totally flat) and there is no interaction between the voice talent. I was amazed to hear that the two 'lead' voices for the Monkey Island series have never actually met (virtually or in real life!), discussed in this great interview by Cressup with Alexandra Boyd who plays Elaine Marley.

These voice over artists are professionals, and usually you never even notice. But unless you have the best setups and the best talent, it's very easy to have video game voice overs that sound disjointed - very different reads of lines stuck together, recorded in obviously different rooms, with little opportunity for dynamism.

Look, I'm not a professional writer (although I want to work with some in the future!), and my dialogue isn't going to naturally 'pop' like something out of a Tarantino movie. So if it's going to come alive, it needs to take from the skills of the actors/actresses who can inhabit a character, and express what is happening to them just through their voice.

I wanted to do something a lot closer to how animated movies at TV shows are made, where they start with the audio performance, and base the final product on it.

One of the references I've been boring the art team with is How to Train Your Dragon, for many reasons (partly as it's one of my 6 year old's favourite movies at the moment), but especially for how much story and character development can be told without dialogue in the animation. But there's a great making-of documentary, that shows how improvisations and different readings of the script made the characters come alive. In this type of production, the animation and editing is done around the audio recording - that's one of the things that come first, after the script and storyboard.

So next week we are trying something really hard. We are going to record the dialogue for the game before a demo or any animation is done, with all the cast in the same room (but in separate glass booths!) so there's a chance to improvise, riff and have true antagonistic interactions between the characters. We've got a cast from stand-up comedy, as well as professional stage and voice over work, but all locally based around Edinburgh so we can do this together.

But our game is not linear. There are hundreds of dialogue choices, several puzzles and multiple endings. Hopefully, the script was written carefully enough that we cover all the different branching dialogue options, and have everything recorded next week we need for the game to make sense! We've also tried to include a bunch of lines that might not be needed - to make a puzzle easier, harder, or for other clues if playtesting comes up with issues!

Then, we will start to make the game around the dialogue, hopefully so that the characters and story will drive the gameplay as much as the puzzles. Voice acting is such an important part of adventure games (and an expensive part), but I think it adds so much to the characters and how much the player is immersed in the game.

Next update we should have some behind the scenes shots from the recording session, as well as first shares of the artwork and, goodness, even the title of this damn game!

Just to make clear again, this game is just a test - it's going to be about a 20 minute game, hopefully out in August 2024, but with all the artwork, puzzles and voice talent we want to have in the Big Game. And if this crazy experiment works, we'll record the dialogue for that in the same way, so that even if it's more planning (and pickups), it feels more like you are listening to a play than just a list of lines on a page.