Back to results
US 2026/0290374 A1Published application
System
AssigneeSoftBank Group Corp.
InventorSeitaro MURA
Life of this patent
Estimating the term…
Abstract
A system includes a processor that is configured to switch a received communication from a human response to an artificial intelligence automatic response, acquire voice data associated with the received communication and convert the voice data into text data by using a speech recognition technique, and analyze the text data by using a natural language processing technique and generate a prompt for determining a falsehood level of a statement contained in the text data.
Claims20
A system comprising:
- circuitry configured to:
- receive, via a communication interface coupled to a packet-switched network, audio stream data associated with a communication session from a terminal device, and switch a handling mode of the communication session from a manual response mode to an automated response mode;
- convert the audio stream data into text data by executing a speech recognition pipeline comprising an acoustic model and a language model;
- execute a natural language processing operation on the text data to extract linguistic feature vectors, and input the linguistic feature vectors into a neural network configured to output a reliability score indicating a confidence level that statements within the text data are consistent and truthful;
- construct an instruction sequence comprising the text data, the reliability score, and a context parameter, and input the instruction sequence into a generative neural network model comprising a transformer-based architecture to generate response action data identifying a recommended action category;
- and
- transmit the response action data to the terminal device via the communication interface.
4 more elements
A system comprising:
- circuitry configured to:
- receive audio stream data associated with a communication session from a terminal device via a communication interface and switch a handling mode from manual response to automated response;
- convert the audio stream data into text data using a speech recognition pipeline, extract linguistic feature vectors via natural language processing, and compute a reliability score using a neural network indicating consistency and truthfulness of statements in the text data;
- extract affective state features from the audio stream data, input the affective state features into an emotion classification model to determine an affective state vector, and construct an instruction sequence comprising the text data, the reliability score, and the affective state vector;
- input the instruction sequence into a generative neural network model to generate response action data adjusted based on the affective state vector;
- and
- transmit the response action data to the terminal device.
4 more elements
A method comprising:
- receiving, by circuitry via a communication interface coupled to a packet-switched network, audio stream data associated with a communication session from a terminal device, and switching a handling mode of the communication session from a manual response mode to an automated response mode;
- converting the audio stream data into text data by executing a speech recognition pipeline comprising an acoustic model and a language model;
- executing a natural language processing operation on the text data to extract linguistic feature vectors, and inputting the linguistic feature vectors into a neural network configured to output a reliability score indicating a confidence level that statements within the text data are consistent and truthful;
- constructing an instruction sequence comprising the text data, the reliability score, and a context parameter, and inputting the instruction sequence into a generative neural network model comprising a transformer-based architecture to generate response action data identifying a recommended action category;
- and
- transmitting the response action data to the terminal device via the communication interface.
3 more elements
Dependencies are inferred from the claim text.