公开(公告)号：US20240304174A1

公开(公告)日：2024-09-12

申请号：US18180665

申请日：2023-03-08

Applicant: Sony Interactive Entertainment Inc.

Inventor： Celeste M.B. Bean ,  Landon G. Noss

IPC: G10L13/027 ,  A63F13/215 ,  A63F13/424 ,  G10L13/04 ,  G10L25/84

CPC classification number: G10L13/027 ,  A63F13/215 ,  A63F13/424 ,  G10L13/04 ,  G10L25/84

Abstract: Methods and systems are presented for discerning background voices from ambient noise in a video game player's surroundings, isolating the voices using distinct phonemes within the voices, and running the phonemes through a generative artificial intelligence (AI) model to produce synthesized speech in the video game. The synthesized speech can be spoken by players or non-player characters, modified in age, gender, and other attributes of the avatar renderings. The text for the synthesized speech can be for static scripts or alterations based on gameplay. The player can select from different background voices and apply them to in-game characters and elements as desired.

公开(公告)号：US20240320867A1

公开(公告)日：2024-09-26

申请号：US18186752

申请日：2023-03-20

Applicant: Sony Interactive Entertainment Inc.

Inventor： Celeste M.B. Bean

IPC: G06T11/00 ,  G06F16/532

CPC classification number: G06T11/00 ,  G06F16/532

Abstract: Methods and systems are presented for automatically identifying additional descriptors of an image generated by a text-to-image generator from an initial prompt. The additional descriptors are either incorporated into the initial prompt or made into a new prompt in order to produce another image from the text-to-image generator. The initial prompt and additional descriptors can describe visual features represented in images including content, artistic styles, visual perspectives, and other visible attributes of images. The additional descriptors can be incorporated into the initial prompt by replacing or supplementing existing descriptors. Subsequent images generated by the text-to-image generator can be used to iteratively produce additional descriptors.