Back to results
US 2026/0291884 A1Published application
System
AssigneeSoftBank Group Corp.
InventorHiroaki MASAKI
Life of this patent
Estimating the term…
Abstract
A system includes a processor that is configured to: acquire audio data and convert the audio data into text data using a speech recognition technique; store the converted text data in a recording medium; generate a prompt for instructing a generative AI model to generate a summary based on the stored text data; input the prompt into the generative AI model; and transmit information via a network in order to share the generated summary immediately.
Claims20
A system comprising:
- circuitry configured to:
- acquire, via a communication interface coupled to a packet-switched network, audio data from a terminal device, execute speech recognition processing on the audio data by extracting acoustic feature vectors and decoding the acoustic feature vectors using a neural network acoustic model and a language model to generate character information;
- structure and store the character information in an information storage device in association with attribute information comprising an event identifier, temporal information, and participant information, the character information being stored as indexed records enabling retrieval by event identifier and temporal range;
- dynamically generate a prompt sentence based on at least one of a length and a content of the character information, the prompt sentence comprising an instruction specification defining at least an output format, an output scope, and extraction targets, and transmit the prompt sentence and the character information to a generative neural network model comprising a transformer-based architecture to acquire at least one of summary information and analysis information;
- parse the summary information and the analysis information to extract structured data elements comprising decision data, task item data, and temporal constraint data, and store the structured data elements in the information storage device in association with the event identifier;
- generate notification data comprising the structured data elements formatted as display control information that distinguishes the decision data, the task item data, and the temporal constraint data, and transmit the notification data to one or more terminal devices via the communication interface;
- and
- acquire, from a terminal device, request information comprising an additional prompt sentence defining additional analysis conditions, reacquire the character information from the information storage device, construct a composite prompt sentence combining the additional prompt sentence with system-level instruction data, and transmit the composite prompt sentence to the generative neural network model to acquire updated analysis information.
5 more elements
A system comprising:
- circuitry configured to:
- acquire audio data via a communication interface, execute speech recognition processing on the audio data to generate character information, and store the character information in an information storage device in association with attribute information;
- dynamically generate a prompt sentence based on properties of the character information and transmit the prompt sentence to a generative neural network model to acquire summary information;
- parse the summary information to extract structured data elements comprising decision data, task item data, and temporal constraint data;
- generate notification data comprising the structured data elements and transmit the notification data to a terminal device;
- and
- acquire an additional prompt sentence from the terminal device, reacquire the character information, and transmit a composite prompt sentence to the generative neural network model to acquire updated analysis information.
4 more elements
A method performed by circuitry via a communication interface coupled to a packet-switched network, the method comprising:
- acquiring audio data from a terminal device and executing speech recognition processing on the audio data by extracting acoustic feature vectors and decoding the acoustic feature vectors using a neural network acoustic model to generate character information;
- structuring and storing the character information in an information storage device in association with attribute information comprising an event identifier and temporal information;
- dynamically generating a prompt sentence based on at least one of a length and a content of the character information, the prompt sentence comprising an instruction specification defining an output format and extraction targets, and transmitting the prompt sentence to a generative neural network model comprising a transformer-based architecture to acquire summary information;
- parsing the summary information to extract structured data elements comprising decision data, task item data, and temporal constraint data;
- generating notification data comprising the structured data elements and transmitting the notification data to one or more terminal devices;
- and
- acquiring an additional prompt sentence from a terminal device, reacquiring the character information, and transmitting a composite prompt sentence to the generative neural network model to acquire updated analysis information.
4 more elements
Dependencies are inferred from the claim text.