Tuesday, September 30, 2008

Day Sixteen: About Voices

Posted by Nicolas (permalink)

Among the most popular questions that were asked during the last two weeks, there's the question were people ask us how we did to record more than 400 lines of speech and if we included a bit of humor inside the game. Now that the release is near, I feel that we can talk a bit more about this part of the development.

To be honest, we didn't record these lines ourselves, we used voice synthesis. But don't take us wrong, we didn't used crappy microsoft voice synthesis, or this kind of things. We used a professional tool from a very important company of this sector. Now, you may ask us, why did we used voice synthesis, and not real actors to make the voices of our NPCs. The answer is quite simple, because of the time, the organization and the equipment that it requires.

Firstly, the scripts were never really fixed, they suffered a lot of changes, even until the end of the development (and even now). With real actors I should have gave them the script once it was fixed, and this situation was too long to reach. We could not wait until the end of the development process to record voices because there were still four hundred and fifty lines of dialogue to record.

Secondly, as I am both a student and an employee in real life, I did not have enough time to manage a team of people solely dedicated to voices. As there were four hundred and fifty lines of dialogue and voice synthesis would quickly generate them, this is why I choosed this system. Once the scripts written, I simply had to copy the lines in the software, launch a mass audio processing on the generated sounds and it gave me the dialogues.

Finally, audio recording requires to have quality microphones and sound cards that we do not own. This is also why we choosed voice synthesis. Now, to prove you that these voices are not as crappy as the ones you may have heared in the past, here's a sample taken from the game:

We had to choose between fewer voices, recorded with our voices and our french accent, and this voice synthesis software which was providing really good results. The choice was fairly easy. Sorry if you're a bit disappointed.