Automatically identifying speakers in real-time through media processing with dialog understanding supported by AI techniques

A computer readable storage medium storing a program of instructions executable by a machine to perform a method comprising:

• receiving a plurality of audio streams, an audio stream representing a speech of a participant speaking into a microphone during an online meeting, the plurality of audio streams forming a dialog among a plurality of participants in the online meeting;
• determining a voice characteristic of a voice corresponding to the speech of the participant in the audio stream;
• converting the plurality of audio streams into text and performing a natural language processing to determine content context of the dialog, the natural language processing identifying a name uttered in at least one of the plurality of the audio streams, to associate with the voice in the audio stream based on the determined content context;
• creating a data structure linking the name with the voice characteristic of the voice and storing the data structure in a knowledge base;
• triggering the name to be presented on a user interface associated with the online meeting application;
• responsive to the participant addressing another speaker by said another speaker's identifier with a question, populating the data structure with said another speaker's identifier;
• responsive to detecting another audio in the dialog and determining said another audio is not recognized in the knowledge base and said another audio does not include said another speaker's identifier, inferring, based on at least determining by the natural language processing that said another audio is made in response to the question, that said another audio belongs to said another speaker;
• and
• linking the data structure populated with said another speaker's identifier with voice characteristics associated with said another audio.

A system of automatically identifying speakers in real-time through media processing with dialog understanding, comprising:

• at least one hardware processor;
• a memory device coupled to the at least one hardware processor;
• the at least one hardware processor operable to:
• receive a plurality of audio streams, an audio stream representing a speech of a participant speaking into a microphone during an online meeting, the plurality of audio streams forming a dialog among a plurality of participants in the online meeting;
• determine a voice characteristic of a voice corresponding to the speech of the participant in the audio stream;
• convert the plurality of audio streams into text and perform a natural language processing to determine content context of the dialog, the natural language processing identifying a name uttered in at least one of the plurality of the audio streams, to associate with the voice in the audio stream based on the determined content context;
• create a data structure linking the name with the voice characteristic of the voice and store the data structure on the memory device as part of a knowledge base;
• trigger the name to be presented on a user interface associated with the online meeting application;
• responsive to the participant addressing another speaker by said another speaker's identifier with a question, populate the data structure with said another speaker's identifier;
• responsive to detecting another audio in the dialog and determining said another audio is not recognized in the knowledge base and said another audio does not include said another speaker's identifier, infer, based on at least determining by the natural language processing that said another audio is made in response to the question, that said another audio belongs to said another speaker;
• and
• link the data structure populated with said another speaker's identifier with voice characteristics associated with said another audio.