IP Library › Granted Patent US 12676155
Granted Patent B2
US 12676155 · App. 18/226,097 · Granted Jul 7, 2026

AI-assisted transcript analysis and script generation

Inventors: Helgi Hilmarsson (Corte Madera, CA); Andrew Sung Jae Kim (Cambridge, MA); Bilung Lee (San Jose, CA); Vijay Venkataswamy Parthasarathy (San Jose, CA); Sasank Vemuri (Pflugerville, TX)
Assignee: Zoom Communications, Inc.
G10L17/22G10L17/02
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12676155
App. No.
18/226,097
Granted
Jul 7, 2026
Kind
B2
Abstract

One example method includes receiving, by a virtual conference provider, audio streams from a plurality of participants during a virtual conference; performing speech recognition, by the virtual conference provider, on the received audio streams to generate a transcript of the virtual conference; determining one or more utterances made during the virtual conference; determining, using a first trained machine learning (“ML”) model, one or more sub-topics discussed during the virtual conference based on the one or more utterances; determining, using a second trained machine learning (“ML”) model, one or more topics discussed during the virtual conference based on the one or more sub-topics; determining metadata corresponding to each of the one or more topics, the one or more sub-topics, and the one or more utterances; and generating and providing suggested follow-up topics for a subsequent virtual conference based on the metadata.

Claims (44)

1 . A method comprising:

receiving, by a virtual conference provider, audio streams from a plurality of participants during a virtual conference;

performing speech recognition, by the virtual conference provider, on the received audio streams to generate a transcript of the virtual conference;

determining, based on the transcript, one or more utterances made during the virtual conference;

determining, using a first trained machine learning (“ML”) model, one or more sub-topics discussed during the virtual conference based on the one or more utterances;

determining, using a second trained machine learning (“ML”) model, one or more topics discussed during the virtual conference based on the one or more sub-topics;

determining metadata corresponding to each of the one or more topics, the one or more sub-topics, and the one or more utterances; and

generating and providing suggested follow-up topics for a subsequent virtual conference based on the metadata.

2 . The method of claim 1 , further comprising determining first and second metadata corresponding to at least one utterance, the first metadata indicating a first relationship with a first sub-topic and the second metadata indicating a second relationship with a second sub-topic.

3 . The method of claim 1 , wherein the metadata comprises one or more of a sentiment or an engagement level of one or more of (i) the one or more utterances, (ii) the one or more sub-topics, or (iii) the one or more topics.

4 . The method of claim 1 , wherein the metadata comprises a speaker of a corresponding utterance of the one or more utterances.

5 . The method of claim 1 , wherein the metadata comprises a weight or a confidence score associated with a feature associated with an utterance of the one or more utterances.

6 . The method of claim 1 , wherein generating the suggested follow-up topics is based on a negative sentiment associated with one or more of (i) the one or more utterances, (ii) the one or more sub-topics, or (iii) the one or more topics.

7 . The method of claim 1 , wherein generating the suggested follow-up topics is based on a level of engagement associated with one or more of (i) the one or more utterances, (ii) the one or more sub-topics, or (iii) the one or more topics.

8 . A system comprising:

a communications interface;

a non-transitory computer-readable medium; and

one or more processors configured to execute processor-executable instructions stored in the non-transitory computer-readable medium to:

receive audio streams from a plurality of participants during a virtual conference;

perform speech recognition on the received audio streams to generate a transcript of the virtual conference;

determine, based on the transcript, one or more utterances made during the virtual conference;

determine, using a first trained machine learning (“ML”) model, one or more sub-topics discussed during the virtual conference based on the one or more utterances;

determine, using a second trained machine learning (“ML”) model, one or more topics discussed during the virtual conference based on the one or more sub-topics;

determine metadata corresponding to each of the one or more topics, the one or more sub-topics, and the one or more utterances; and

generate and providing suggested follow-up topics for a subsequent virtual conference based on the metadata.

9 . The system of claim 8 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to determine first and second metadata corresponding to at least one utterance, the first metadata indicating a first relationship with a first sub-topic and the second metadata indicating a second relationship with a second sub-topic.

10 . The system of claim 8 , wherein the metadata comprises one or more of a sentiment or an engagement level of one or more of (i) the one or more utterances, (ii) the one or more sub-topics, or (iii) the one or more topics.

11 . The system of claim 8 , wherein the metadata comprises a speaker of a corresponding utterance of the one or more utterances.

12 . The system of claim 8 , wherein the metadata comprises a weight or a confidence score associated with a feature associated with an utterance of the one or more utterances.

13 . The system of claim 8 , wherein generating the suggested follow-up topics is based on a negative sentiment associated with one or more of (i) the one or more utterances, (ii) the one or more sub-topics, or (iii) the one or more topics.

14 . The system of claim 8 , wherein generating the suggested follow-up topics is based on a level of engagement associated with one or more of (i) the one or more utterances, (ii) the one or more sub-topics, or (iii) the one or more topics.

15 . A non-transitory computer-readable medium comprising processor-executable instructions configured to cause one or more processors to:

receive audio streams from a plurality of participants during a virtual conference;

perform speech recognition on the received audio streams to generate a transcript of the virtual conference;

determine, based on the transcript, one or more utterances made during the virtual conference;

determine, using a first trained machine learning (“ML”) model, one or more sub-topics discussed during the virtual conference based on the one or more utterances;

determine, using a second trained machine learning (“ML”) model, one or more topics discussed during the virtual conference based on the one or more sub-topics;

determine metadata corresponding to each of the one or more topics, the one or more sub-topics, and the one or more utterances; and

generate and providing suggested follow-up topics for a subsequent virtual conference based on the metadata.

16 . The non-transitory computer-readable medium of claim 15 , wherein the metadata comprises one or more of a sentiment or an engagement level of one or more of (i) the one or more utterances, (ii) the one or more sub-topics, or (iii) the one or more topics.

17 . The non-transitory computer-readable medium of claim 15 , wherein the metadata comprises a speaker of a corresponding utterance of the one or more utterances.

18 . The non-transitory computer-readable medium of claim 15 , wherein the metadata comprises a weight or a confidence score associated with a feature associated with an utterance of the one or more utterances.

19 . The non-transitory computer-readable medium of claim 15 , wherein generating the suggested follow-up topics is based on a negative sentiment associated with one or more of (i) the one or more utterances, (ii) the one or more sub-topics, or (iii) the one or more topics.

20 . The non-transitory computer-readable medium of claim 15 , wherein generating the suggested follow-up topics is based on a level of engagement associated with one or more of (i) the one or more utterances, (ii) the one or more sub-topics, or (iii) the one or more topics.