Abstract: Technology is disclosed for programmatically determining, for a segment of a meeting recording, a user-specific adaptive playback speed, and generating a time-stretched segment playable at the adaptive playback speed. The adaptive playback speed is faster or slower than a default playback speed of the meeting recording. To determine the adaptive playback speed, this disclosure provides technologies to determine a playback data feature based on user-meeting data. The adaptive playback is generated based on the playback data feature. The segment is time-stretched to the adaptive playback speed to generate an updated meeting recording including the segment that is time-stretched and playable at the adaptive playback speed. In this manner, an updated meeting recording, specific to a user, and playable at an adaptive playback speed based on user-meeting data may reduce bandwidth associated with user's manually editing videos or rewinding playback, while improving user experience.

Abstract: Technology is disclosed for programmatically determining, for a segment of a meeting recording, a user-specific adaptive playback speed, and generating a time-stretched segment playable at the adaptive playback speed. The adaptive playback speed is faster or slower than a default playback speed of the meeting recording. To determine the adaptive playback speed, this disclosure provides technologies to determine a playback data feature based on user-meeting data. The adaptive playback is generated based on the playback data feature. The segment is time-stretched to the adaptive playback speed to generate an updated meeting recording including the segment that is time-stretched and playable at the adaptive playback speed. In this manner, an updated meeting recording, specific to a user, and playable at an adaptive playback speed based on user-meeting data may reduce bandwidth associated with user's manually editing videos or rewinding playback, while improving user experience.

Abstract: Various embodiments discussed herein are directed to improving existing technologies by detecting a likelihood of hallucination arising from one-shot, few-shot, or outside knowledge contexts. For example, regarding the one-shot or few-shot contexts, some embodiments determine a set of tokens in a language model output that are not found in target content but are found in at least one example. When such phrases are not very common words, this is highly indicative that the model is hallucinating because these phrases should be located in the target content but are not, but are instead located in the examples.

Abstract: Technology is disclosed to programmatically generate a briefing message that is relevant, optimally-sized, and easy for a user to consume. The briefing message is generated with target briefing items organized based on a user-specific headline and user-specific body of the briefing message. The briefing message is generated based on a determination, from candidate briefing items and corresponding scores, of at least one target briefing item to be included in the body and the headline of a briefing message. In this manner, such technologies improve the user's experience and reduce computational resources and bandwidth associated with a larger volume of content being generated and communicated to a user for presentation, especially during times of high user computer traffic. The briefing content items are generated during times of the day with lower user activity and less computational resource consumption.

Abstract: Various embodiments discussed herein are directed to improving existing technologies by providing a corpus data supplement as input into a model, such as a Large Language Model (LLM). Consequently, the model can generate accurate scores or data for predictions because the model is better able to distinguish between a general understanding of natural language concepts and domain-specific concepts.

Abstract: In some implementations, a method includes extracting completion criteria of an action item and parameters of the completion criteria from a message portion of a user message between users. In response to determining, from sensor data provided by one or more sensors associated with a user, the user practices a routine; a routine-related aspect is generated from a user routine model of the user for the routine. In is inferred at least one of the extracted completion criteria of the action item is unsatisfied based on the extracted parameters and the identified routine-related aspect. Based on the inferring, a notification is provided to at least one user associated with the action item.

Abstract: Systems and methods for surfacing natural language queries from one or more transcripts. An example method may include converting received audio to text, through automated speech recognition, to form a transcript of the audio, wherein the transcript includes text of the audio and identifications of speakers associated with portions of the text corresponding to utterances from the respective speakers; generating input signals based on at least the transcript; executing at least one of one or more heuristics or a trained machine-learning (ML) model, using the generated input signals as an input, to generate at least one of a suggested natural language query for searching the transcript or a key moment within the received audio; and causing at least one of the suggested natural language query or the key moment to be surfaced on one or more remote devices.

Abstract: Technology is disclosed for programmatically determining, for a segment of a meeting recording, a user-specific adaptive playback speed, and generating a time-stretched segment playable at the adaptive playback speed. The adaptive playback speed is faster or slower than a default playback speed of the meeting recording. To determine the adaptive playback speed, this disclosure provides technologies to determine a playback data feature based on user-meeting data. The adaptive playback is generated based on the playback data feature. The segment is time-stretched to the adaptive playback speed to generate an updated meeting recording including the segment that is time-stretched and playable at the adaptive playback speed. In this manner, an updated meeting recording, specific to a user, and playable at an adaptive playback speed based on user-meeting data may reduce bandwidth associated with user's manually editing videos or rewinding playback, while improving user experience.

Abstract: Systems and methods for surfacing natural language queries from one or more transcripts. An example method may include converting received audio to text, through automated speech recognition, to form a transcript of the audio, wherein the transcript includes text of the audio and identifications of speakers associated with portions of the text corresponding to utterances from the respective speakers; generating input signals based on at least the transcript; executing at least one of one or more heuristics or a trained machine-learning (ML) model, using the generated input signals as an input, to generate at least one of a suggested natural language query for searching the transcript or a key moment within the received audio; and causing at least one of the suggested natural language query or the key moment to be surfaced on one or more remote devices.

Abstract: Intelligent agents (IA) for automatically generating responses to content within a communication session (CS) are disclosed. An IA is trained to target the responses to a user and the user's context within the CS. An IA receives CS content that includes natural language expressions encoding users' conversations and determines content features based on natural language models. The content features indicate intended semantics of the expressions. The IA identifies likely-relevant content to the targeted user, to generate a response for. Identifying such content includes determining a relevance of the content based on content features, a context of the CS, a user-interest model, and a content-relevance model. Identifying the likely-relevant content to respond to is based on the determined relevance of the content and relevance thresholds. Various responses to the identified portions of the content are automatically generated and provided based on a natural language response-generation model targeted to the user.

Abstract: Personalized information from venues of interest to a user is provided to the user. The information may comprise messages, notifications, or other information determined to be relevant to the user, and may be timely provided to the user in an appropriate format such that the user is better able to respond to the information. A set of venues that are relevant to the user and corresponding information sources associated with the venue are determined. Information published by these sources may be analyzed for relevance to the user, and the relevant information items may be provided to the user via notifications or applications and services. Further, the notifications may be scheduled for presentation to the user at appropriate times, such as prior to a predicted visit of the venue. One embodiment includes a venue-information clearinghouse for receiving information from venues and making it available for user applications or services.

Abstract: Intelligent agents (IA) for automatically generating responses to content within a communication session (CS) are disclosed. An IA is trained to target the responses to a user and the user's context within the CS. An IA receives CS content that includes natural language expressions encoding users' conversations and determines content features based on natural language models. The content features indicate intended semantics of the expressions. The IA identifies likely-relevant content to the targeted user, to generate a response for. Identifying such content includes determining a relevance of the content based on content features, a context of the CS, a user-interest model, and a content-relevance model. Identifying the likely-relevant content to respond to is based on the determined relevance of the content and relevance thresholds. Various responses to the identified portions of the content are automatically generated and provided based on a natural language response-generation model targeted to the user.

Abstract: Technology is disclosed for improving user privacy and providing user control over user-activity data collected from personal computing devices (i.e., user devices). User devices may be configured to operate in a private mode that enables a user to control, for example, which aspects of user-activity data are provided to applications and services running on their user device; to obscure or modify aspects of user-activity data so that certain applications and services, which may require this information to operate, may still function, but that the obscured information provided to these applications and services preserves user privacy or no longer may be used to identify the user; or to remove evidence of user-activity data created, monitored, reported, or otherwise collected by or on the user device while the user is operating their user device in the private mode setting.

Abstract: Techniques of streaming data processing for hybrid online meetings are disclosed herein. In one example, a method includes receiving, at the remote server, a video stream captured by a camera in the conference room. The video stream captures images of multiple local participants of an online meeting. The method also includes determining identities of the captured images of the multiple local participants in the received video stream using meeting information of the online meeting and generating a set of individual video streams each corresponding to one of the multiple local participants. The set of individual video streams can then be transmitted to the second computing device corresponding to a remote participant of the online meeting as if the multiple local participants are virtually joining the online meeting.

Abstract: Methods for providing enhanced services to users participating in communication sessions (CS), via a virtual assistant, are disclosed. One method receives content that is exchanged by users participating in the CS. The content includes natural language expressions that encode a conversation carried out by users. The method determines content features based on natural language models. The content features indicate intended semantics of the natural language expressions. The method determines a relevance of the content and identifies portions of the content that are likely relevant to the user. Determining the relevance is based on the content features, a context of the CS, a user-interest model, and a content-relevance model of the natural language models. Identifying the likely relevant content is based on the determined relevance of the content and a relevance threshold. A summary of the CS is automatically generated from summarized versions of the likely relevant portions of the content.

Abstract: Various embodiments discussed herein are directed to improving existing technologies by causing certain characters to be replaced at a document if such characters are likely to be an error. For example, documents generated using speech-to-text technology or Optical Character Recognition (OCR) technology often contain character errors. A scoring threshold may be utilized to determine one or more characters are not being correctly represented in the document. Alternatively or additionally, various embodiments recommend multiple character sequences as candidates to replace other characters and a user may select which of the candidates will be used for replacement.

Abstract: Systems and methods for surfacing natural language queries from one or more transcripts. An example method may include converting received audio to text, through automated speech recognition, to form a transcript of the audio, wherein the transcript includes text of the audio and identifications of speakers associated with portions of the text corresponding to utterances from the respective speakers; generating input signals based on at least the transcript; executing at least one of one or more heuristics or a trained machine-learning (ML) model, using the generated input signals as an input, to generate at least one of a suggested natural language query for searching the transcript or a key moment within the received audio; and causing at least one of the suggested natural language query or the key moment to be surfaced on one or more remote devices.

Abstract: Techniques of streaming data processing for hybrid online meetings are disclosed herein. In one example, a method includes receiving, at the remote server, a video stream captured by a camera in the conference room. The video stream captures images of multiple local participants of an online meeting. The method also includes determining identities of the captured images of the multiple local participants in the received video stream using meeting information of the online meeting and generating a set of individual video streams each corresponding to one of the multiple local participants. The set of individual video streams can then be transmitted to the second computing device corresponding to a remote participant of the online meeting as if the multiple local participants are virtually joining the online meeting.

Abstract: Techniques of streaming data processing for hybrid online meetings are disclosed herein. In one example, a method includes receiving, at the remote server, a video stream captured by a camera in the conference room. The video stream captures images of multiple local participants of an online meeting. The method also includes determining identities of the captured images of the multiple local participants in the received video stream using meeting information of the online meeting and generating a set of individual video streams each corresponding to one of the multiple local participants. The set of individual video streams can then be transmitted to the second computing device corresponding to a remote participant of the online meeting as if the multiple local participants are virtually joining the online meeting.

Abstract: In some implementations, a method includes extracting message attributes of an email associated with a user from the email. User interaction data is identified that is generated by the user in association with display of the email based on sensor data from one or more sensors. It is determined that the user interaction data corresponds to a routine of the user based on a routine-related aspect generated from a user routine model representing the routine. A time to present a notification of the email is determined based on the routine. The notification is provided to the user on a user device based on the determined time to present the notification.