System

A system comprising:

• a communication interface coupled to a packet-switched network;
• and
• circuitry configured to:
• acquire, via the communication interface, a digital audio signal representing a voice input of a user from a terminal device;
• extract a sequence of acoustic feature vectors from the digital audio signal and input the sequence into a speech recognition neural network model to generate text data;
• analyze the text data using a natural language processing operation to extract intent information and entity information;
• acquire sensor data via the communication interface and integrate the sensor data with the intent information and the entity information to generate context information;
• estimate an emotional state of the user based on voice feature data extracted from the digital audio signal;
• generate a prompt sentence based on the intent information, the context information, and the estimated emotional state, the prompt sentence being configured to instruct a generative neural network model comprising a transformer architecture to generate an instruction output;
• input the prompt sentence to the generative neural network model and obtain output text data;
• verify the output text data against a set of stored constraint rules and, when the output text data fails to satisfy at least one constraint rule, regenerate the output text data using a modified prompt sentence;
• segment verified output text data into a plurality of unit segments and associate each unit segment with a corresponding reference point;
• convert each unit segment into audio data using a speech synthesis process;
• and
• selectively transmit the audio data to the terminal device based on position data received from the terminal device.

A method comprising:

• acquiring, via a communication interface coupled to a packet-switched network, a digital audio signal representing a voice input of a user from a terminal device;
• executing a speech recognition process on the digital audio signal to generate text data;
• analyzing the text data together with context information to generate intent information representing an intention of the user;
• generating a prompt sentence based on the intent information and the context information, the prompt sentence being configured to instruct a generative neural network model to generate an instruction output;
• inputting the prompt sentence to the generative neural network model and obtaining, from the generative neural network model, output text data representing the instruction output corresponding to the intent information;
• performing a speech synthesis process on the output text data to convert the output text data into audio data;
• and
• transmitting the audio data to the terminal device via the communication interface.