Abstract: System, process and device configurations are provided for detecting negative gameplay behavior and gameplay control. A method can include receiving gameplay data for a plurality of players for a game, wherein the gameplay data includes player input controls for the plurality of players. The process may use player input controls to detect negative gameplay behavior, such as a toxic behavior, including but not limited to a team kill, item/equipment trade fraud. Player input controls may be identified and evaluated using a model. Processes can include warning players based on patterns observed by the model and blocking toxic player input controls meant to harm other players. Processes can include determining game function responses of the game to limit outcome of the at least one player input control for the game and controlling output of the game.

Abstract: Techniques are provided for enabling game developers to create speech from text. The pitch of intermediate representations of phonemes is tailored on a phoneme-by-phoneme basis from the pitch output by a text-to-speech model.

Abstract: A machine learning (ML) model is used to automatically generate haptics signals to actuate a haptics generator in a computer game controller. The haptics signal is generated based on audio from the game input to the ML model. Current controller operation and other parameters also may be input to the M model to modify the haptics signal. Category importance and frequency may be applied to the loss function of the ML model to further refine haptics generation. Post-filtering may be used to reduce false positives. Game genre may be used to reduce the number of candidate haptics signals for generation.

Abstract: System, process and device configurations are provided for detecting negative gameplay behavior and gameplay control. A method can include receiving gameplay data for a plurality of players for a game, wherein the gameplay data includes player input controls for the plurality of players. The process may use player input controls to detect negative gameplay behavior, such as a toxic behavior, including but not limited to a team kill, item/equipment trade fraud. Player input controls may be identified and evaluated using a model. Processes can include warning players based on patterns observed by the model and blocking toxic player input controls meant to harm other players. Processes can include determining game function responses of the game to limit outcome of the at least one player input control for the game and controlling output of the game.

Abstract: Systems, processes and device configurations are provided for electronic gaming control and performance based sound generation for network games, such as multiplayer or cloud based games. Processed include determining one or more sound cues for a game data stream. By generating and outputting one or more sound cues on the fly, sound feedback may be provided to compliment gameplay and a game experience. Sound cues can transform or modify audio elements of games for users and viewers of electronic gaming content. Generating sound cues may include replacing one or more assets of an electronic game, and capturing the user interaction with the modified content. Generating sound cues may include generating one or more sounds for a user, such as crowd cheers or boos, wherein the underlying game code and game content is not modified. Processes and configurations are also provided for determining sound assets of gaming content and tracking user performance.

Abstract: System, process and device configurations are provided for generating game commentary streams. Game commentary streams can include gameplay video data with voice data of multiplayer game users synchronized to one or more game events and game states in recorded gameplay. Game commentary streams may allow for selection of game play content from the perspective of a selected user and to obtain voice data of the user and one or more other users during the game session. Methods for generating a commentary stream can include receiving game commentary data from users of a multiplayer electronic game including voice data detected during the game session and generating the commentary stream to include gameplay data for the game session and selected game commentary data. User requests may be received and the commentary stream may be output for recorded game play from the perspective of one or more users of the game session.

Abstract: Methods and systems are provided for selecting a viewport into a video game for a spectator viewing a video game played by a player. The method includes identifying a plurality of virtual cameras for providing a plurality of viewports into scenes of the video game. The method includes accessing a spectator playbook, the spectator playbook is stored in association with a profile of the spectator. The spectator playbook identifies viewing history of the spectator for types of video games viewed by the spectator. The method includes accessing a player playbook, the player playbook is stored in association with a profile of the player. The player playbook identifies performance of the player for types of video games played by the player. The method includes accessing gameplay data for the video game played by the player. The method includes selecting a viewport from the plurality of viewports into the video game.

Abstract: Methods and systems for generating social posts for users watching media content include processing inputs provided by one or more users of a group viewing the media content. Mood of each user in the group is predicted based on context of a portion of the media content currently rendering and the inputs of the users. A social post is generated for each user based on the predicted mood of the user. The social post is forwarded for rendering as input from the user.

Abstract: Methods and systems for generating social posts for users watching media content include processing inputs provided by one or more users of a group viewing the media content. Mood of each user in the group is predicted based on context of a portion of the media content currently rendering and the inputs of the users. A social post is generated for each user based on the predicted mood of the user. The social post is forwarded for rendering as input from the user.

Abstract: Methods and systems are provided for selecting a viewport into a video game for a spectator viewing a video game played by a player. The method includes identifying a plurality of virtual cameras for providing a plurality of viewports into scenes of the video game. The method includes accessing a spectator playbook, the spectator playbook is stored in association with a profile of the spectator. The spectator playbook identifies viewing history of the spectator for types of video games viewed by the spectator. The method includes accessing a player playbook, the player playbook is stored in association with a profile of the player. The player playbook identifies performance of the player for types of video games played by the player. The method includes accessing gameplay data for the video game played by the player. The method includes selecting a viewport from the plurality of viewports into the video game.