System

A system comprising:

• circuitry configured to:
• receive, via a communication interface coupled to a packet-switched network, audio data from a plurality of terminal devices in a communication session and automatically determine a source language for each audio stream;
• perform speech recognition on the audio data using acoustic models and language models corresponding to the determined source languages to generate text data;
• translate the text data into target languages using a machine translation model and refine the translated text data by transmitting a prompt data structure to a generative neural network model comprising a transformer architecture with a plurality of self-attention layers;
• convert the refined text data into synthesized audio data using a speech synthesis model with prosodic characteristics matched to source speakers;
• store the text data and the translated text data as transcript data in a storage device and generate summary data and action item data from the transcript data using the generative neural network model;
• and
• transmit the synthesized audio data, transcript data, and summary data to the respective terminal devices via the communication interface.

A method performed by circuitry of a server coupled to a packet-switched network via a communication interface, the method comprising:

• receiving, via the communication interface, audio data from a plurality of terminal devices in a communication session;
• performing speech recognition on the audio data to generate text data for each terminal device;
• translating the text data into one or more target languages associated with respective terminal devices;
• converting the translated text data into synthesized audio data;
• and
• transmitting the synthesized audio data to the respective terminal devices via the communication interface.