|
|
 |
It all starts with the voice creation process. This works much like a normal recording - you start with a script, and recordings from a voice actor. No special recordings or script are required. These are fed into PA Studio, which performs extensive acoustic and linguistic analysis. The result is a voice file, which is a complete representation of the actor's voice. |
| The voice file is used by Composer which allows you to intelligently combine the recordings to make completely new sentences. The new sentences sound perfectly natural and so composer provides a way of generating millions of recordings from a few hundred or thousand lines of recorded dialog. |
| Generator can say literally anything - its input is free text. It uses the voice file from the voice creation process to generate speech which is a faithful impression of the original actor. The original recordings are not needed, and so Generator can create speech with minimal memory requirements. |
| Both Composer and Generator can be included in the game engine. With Composer, the develop creates compiled composer files offline in PA studio, and uses to the game AI to select these as needed. Composer engine then efficiently combines the recordings, stored on disc or DVD to create the final speech. |
| |
| When used in a game, generator includes the voice file itself. The game AI, or indeed, player, sends text into Generator, which then converts this into speech. |
|
| |
|
|
|
