Transition-driven transcript search

Patent metadata

Field	Value
Publication number	US-12158902-B2
Application number	US-202117305976-A
Country	US
Kind code	B2
Filing date	Jul 19, 2021
Priority date	Nov 18, 2020
Publication date	Dec 3, 2024
Grant date	Dec 3, 2024

Abstract

Methods, systems, and computer programs are presented for searching the content of voice conversations. The conversations are translated into text and analysis of the conversation is performed to identify information in the conversation. The information identified includes turns taking data in the conversation and states identified within each state. A powerful user interface (UI) is provided to review the conversations and add annotations that tag the different turns. Additionally, parameter values are extracted from the text. A powerful search engine is provided with multiple search options, such as searching for text, searching by state within the conversation, searching by parameters extracted from the conversation, or a combination thereof.

First claim

The invention claimed is:

1. A method comprising:
identifying, by a machine-learning (ML) model, portion topics with corresponding portion parameters and corresponding portion outcomes in one or more transcripts, the ML model having been trained with data from a training set that comprises training transcripts with identified portion topics with corresponding portion parameters and corresponding portion outcomes within the training transcripts;
accessing, by one or more processors, transcript data that includes text of multiple transcript portions from the one or more transcripts, each transcript portion among the multiple transcript portions having a corresponding portion topic that is detailed by a corresponding portion parameter and detailed by a corresponding portion outcome;
providing, by the one or more processors, a user interface (UI) for searching the transcript data based on a specified conjunction of a specified portion topic with at least one of a specified portion parameter or a specified portion outcome;
performing, by the one or more processors, a search for one or more transcript portions among the multiple transcript portions based on the specified conjunction of the specified portion topic with at least one of the specified portion parameter or the specified portion outcome; and
causing presentation of results from the performed search within the UI.

2. The method of claim 1 , wherein features of the ML model include one or more turns within the one or more transcripts.

3. The method of claim 1 , wherein performing the search further comprises:
converting a single search query into multiple searches; and
combining results of the multiple searches into one output.

4. The method of claim 1 , wherein the UI includes a search option operable to specify a first value for a first state tag, wherein performing the search further comprises:
identifying one or more transcript portions that include the first value for the first state tag.

5. The method of claim 1 , further comprising:
identifying turns within the transcript data.

6. The method of claim 1 , wherein the UI includes a search option operable to specify at least two state tags combined using at least one of an AND comparison operator or an OR comparison operator.

7. The method of claim 6 , wherein the UI includes a search option operable to exclude a second state tag from the search.

8. The method of claim 1 , further comprising:
creating the transcript data, the creating the transcript data comprising:
obtaining recordings of audio conversations;
performing natural language processing on the recordings to generate raw text without punctation marks and without proper capitalization; and
formatting, by an annotator ML model, the raw text of the audio conversations to generate formatted text.

9. The method of claim 1 , wherein the UI includes a search option operable to specify metadata comprising at least one of duration or names of parties.

10. A system, comprising:
one or more memories; and
one or more processors, communicatively coupled to the one or more memories, configured to perform operations comprising:
identifying, by a machine-learning (ML) model, portion topics with corresponding portion parameters and corresponding portion outcomes in one or more transcripts, the ML model having been trained with data from a training set that comprises training transcripts with identified portion topics with corresponding portion parameters and corresponding portion outcomes within the training transcripts;
accessing transcript data that includes text of multiple transcript portions from the one or more transcripts, each transcript portion among the multiple transcript portions having a corresponding portion topic that is detailed by a corresponding portion parameter and detailed by a corresponding portion outcome;
providing a user interface (UI) for searching the transcript data based on a specified conjunction of a specified portion topic with at least one of a specified portion parameter or a specified portion outcome;
performing a search for one or more transcript portions among the multiple transcript portions based on the specified conjunction of the specified portion topic with at least one of the specified portion parameter or the specified portion outcome; and
causing presentation of results from the performed search within the UI.

11. The system of claim 10 , wherein features of the ML model include one or more turns within the one or more transcripts.

12. The system of claim 10 , wherein the one or more processors are further configured to:
convert a single search query into multiple searches outputting multiple search results; and
combine the multiple search results into a single output.

13. The system of claim 10 , wherein the UI includes a search option operable to specify at least two state tags combined using at least one of an AND comparison operator or an OR comparison operator.

14. A non-transitory computer-readable medium storing a set of instructions, the set of instructions comprising:
one or more instructions that, when executed by one or more processors of a computing device, cause the computing device to perform operations comprising:
identifying, by a machine-learning (ML) model, portion topics with corresponding portion parameters and corresponding portion outcomes in one or more transcripts, the ML model having been trained with data from a training set that comprises training transcripts with identified portion topics with corresponding portion parameters and corresponding portion outcomes within the training transcripts;
accessing transcript data that includes text of multiple transcript portions from the one or more transcripts, each transcript portion among the multiple transcript portions having a corresponding portion topic that is detailed by a corresponding portion parameter and detailed by a corresponding portion outcome;
providing a user interface (UI) for searching the transcript data based on a specified conjunction of a specified portion topic with at least one of a specified portion parameter or a specified portion outcome;
performing a search for one or more transcript portions among the multiple transcript portions based on the specified conjunction of the specified portion topic with at least one of the specified portion parameter or the specified portion outcome; and
causing presentation of results from the performed search within the UI.

15. The non-transitory computer-readable medium of claim 14 , wherein features of the ML model include one or more turns within the one or more transcripts.

16. The non-transitory computer-readable medium of claim 14 , wherein performing the search further comprises:
converting a single search query into multiple searches outputting multiple search results; and
combining the multiple search results into one output.

17. The non-transitory computer-readable medium of claim 14 , wherein the UI includes a search option operable to specify at least two state tags combined using at least one of an AND comparison operator or an OR comparison operator.

Assignees

• Twilio Inc

Inventors

• De Oliveira Luke Percival
• Akeel Umair
• Rodrigo Alfredo Láinez
• Amador Nicolas Acosta
• KUMAR Sahil
• Dremer Liat Barda
• Ahn Byeongung
• Cole Tyler

Classifications

• G06F40/295

  Named entity recognition · CPC title

• G10L15/26

  Speech to text systems (G10L15/08 takes precedence) · CPC title

• G06F16/383

  using metadata automatically derived from the content · CPC title

• G06N20/00

  Machine learning · CPC title

• G06F40/169 Primary

  Annotation, e.g. comment data or footnotes · CPC title