A new patent by EA has been uncovered, which will bring a whole new way of giving gamers an immersive experience. The patent by Electronic Arts will give gamers a way to add their voice to a player in a video game. The patent is dated October 17th and explains the complete process, including 10 drawings, specifications, and claims.
Patents aren’t required to mention any projects where they are planned to be used, and so is the case with this patent as well, but the future sounds quite interesting for Electronic Arts.
Generating Speech in a Player’s Voice Is EA’s New Patent
The patent explains that previous attempts at “generating speech audio in a particular voice” involved hours of speech samples to help the voice synthesizer “accurately capture the voice in the synthesized speech audio.” The patent is said to generate speech audio after capturing not just the player’s voice but also other elements like tone, emotions, and emphasis.
The voice converter, said to be the most vital part of the mechanism, will be configured to initially input “a target speaker embedding,” which will be associated with the player’s voice and the representation that will be continually trained.
Following this, the source acoustic features will reportedly generate one or more acoustic feature encodings, which will be fed to the acoustic decoder to output target acoustic features using one or more vocoders to generate speech in the target voice.
Apart from capturing and generating player voices, the patent is also said to allow gamers to type the required dialog for characters in the game to deliver. This would make generating unique dialogs a lot simpler compared to training the tech to capture and regenerate voice after capturing enough of the player’s voice. Text-to-speech would be a far simpler approach, but it wouldn’t be the same as the experience of listening to one’s voice being delivered by a character in the game while playing.
Electronic Arts Also Patented Tech to Make Video Game Characters Age Appropriate
A few months ago, another patent filed by Electronic Arts was uncovered that was aimed at changing video game characters as they age in the game. The patent is named Voice Aging Using Machine Learning and describes systems and methods in place to achieve ageing voice audio.
The method is said to make use of machine learning to train and convert an input audio signal and an age embedding. The age embedding is said to be based on “an age classification of a plurality of speech audio samples of subjects in a target age category.”
The 2022 title Sifu is a good example of a character aging as the game progresses, and in cases like these, the aging voice process would ensure the age-appropriate voice is showcased perfectly without the need for another voice artist for the same character to get that perfect voice.
A text-to-speech system is said to assist in achieving this target, according to the patent. The text-to-speech system is said to generate an audio signal, which is then sent to what’s called an “age convertor model” to complete the transformation of voice to suit the character’s age.
The two patents by Electronic Arts might just predict what’s to come in the future, not just from EA but also from the entire gaming industry. A story-telling of a character from a young age where the player’s voice gets used to a mature or old age even as the game progresses where the voice-aging mechanism comes into play might take the authenticity to another level.
Patents might not always mean the said technology is ready to use, as it might still be a minute before it’s seen in action in an Electronic Arts game.
Follow us for more entertainment coverage on Facebook, Twitter, Instagram, and YouTube.
