EA working on player-voiced characters in games, patent shows
A patent filed by EA reveals the company is looking into ways to implement player-voiced characters into its games.
The patent, which was spotted by veryaligaming and reshared on social media by NextGenPlayer, aims to allow players to make their in-game characters communicate by using their own voices.
The patent was submitted back in 2020, but was recently made available to view publicaly.
«A computer-implemented method of generating speech audio in a video game is provided. The method includes inputting, into a synthesiser module, input data that represents speech content,» the patent's abstract reads.
Source acoustic features for the speech content in the voice of a source speaker are generated and are input, along with a speaker embedding associated with a player of the video game into an acoustic feature encoder of a voice converter
«One or more acoustic feature encodings are generated as output of the acoustic feature encoder, which are inputted into an acoustic feature decoder of the voice converter to generate target acoustic features. The target acoustic features are processed with one or more modules, to generate speech audio in the voice of the player.»
You can see an example of this method and its application in the image below.
«Many modern video games provide players with the ability to create their own personalised characters, or avatars,» EA writes towards the end of its patent.
«In some video games, players may be able to configure their avatars to speak in different voices. However, previous approaches of generating speech audio in a particular voice typically require many speech samples (e.g. hours of speech samples) in order for the synthesiser to accurately capture the voice in the synthesised
Read more on eurogamer.net
