# US 11,114,092 B2 · Real-time voice processing systems and methods

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 11,114,092 B2
- Document ID: US11114092B2
- Status: Granted
- Kind: B2
- Application number: 16/845,300
- Filed: 2020-04-10
- Published: 2020-08-13
- Granted: 2021-09-07
- Earliest claimed: 2019-02-11
- Assignees: Groupe Allo Media SAS
- Applicants: Groupe Allo Media SAS
- Inventors: Romain Sambarino
- Web page: https://patentgrep.com/patent/US11114092B2

## CPC

- G10L 15/183: using context dependencies, e.g. language models
- G10L 15/26: Speech to text systems (G10L15/08 takes precedence)
- G10L 15/32: Multiple recognisers used in sequence or in parallel; Score combination systems therefor, e.g. voting systems
- H04M 3/5175: {Call or contact centers supervision arrangements}
- H04M 2201/40: using speech recognition
- H04M 2203/252: where a voice mode is enhanced with visual information
- H04M 2203/552: Call annotations
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- H04M: TELEPHONIC COMMUNICATION (circuits for controlling other apparatus via a telephone cable and not involving telephone switching apparatus G08)

## Abstract

A computer-implemented method and supporting system transcribes spoken words being monitored from a telephonic interaction among two or more individuals. Telephonic interactions among the individuals are monitored, and at least two of the individuals are each assigned to a separate channel. While still being monitored, each of the channels is assigned a context-based speech recognition models, and in substantially real-time, the monitored telephonic interaction is transcribed from speech to text based on the different assigned models.

## Claims

### Claim 1

A computer-implemented method for protecting access to
information while transcribing spoken words to text, the method comprising:
electronically monitoring a telephonic interaction, the interaction comprising at least two different channels;
while monitoring the telephonic interaction, assigning one of a plurality of context-based speech recognition models to at least one of the two different channels wherein the one of the plurality of context-based speech recognition models is trained to recognize protected information and the plurality of context-based speech recognition models are organized in a multi-level hierarchical structure;
for at least one of the two different channels, swapping the assigned context-based speech recognition model with an alternate context-based speech recognition model while still monitoring the telephonic interaction and, wherein the speech from one of the other of the two different channels comprises a question, and wherein the alternate context-based speech recognition model comprises a fixed set of possible answers for the question; and
as the text is being transcribed, transcribing the monitored telephonic interaction from speech to text based on the assigned model such that the protected information is obfuscated.

### Claim 2 (depends on 1)

The method of claim 1, wherein one of the two different channels represents a service consumer and the other one of the two different channels represents a service provider.

### Claim 3 (depends on 1)

The method of claim 1, wherein the protected information is subject to legal protections under one or more statutes.

### Claim 4 (depends on 3)

The method of claim 3, wherein the one or more statutes comprise the General Data Protection Regulation (“GDPR”), the Health Insurance Portability and Accountability Act (“HIPAA”), or the Family Educational Rights and Privacy Act (“FERPA”).

### Claim 5 (depends on 1)

The method of claim 1, wherein a subset of the plurality of context-based speech recognition models include role-specific variants such that the assignment of one of the plurality of context-based speech recognition models to one of the channels is based in part on a transaction role of one of the channels.

### Claim 6 (depends on 1)

The method of claim 1, wherein the protected information comprises financial account data.

### Claim 7 (depends on 1)

The method of claim 1, wherein the protected information comprises healthcare data.

### Claim 8 (depends on 1)

The method of claim 1, wherein the alternate context-based speech recognition model assigned to at least one of the two different channels is based at least in part on speech from one of the other of the two different channels.

### Claim 9 (depends on 1)

The method of claim 1, further comprising presenting, via an electronic user interface and while monitoring the telephonic interaction, the transcribed text to a user operating as one of the channels such that the protected information cannot be seen by the user.

### Claim 10

A system for protecting information while transcribing spoken words to text, the system comprising:
at least one memory for storing computer-executable instructions; and
at least one processor for executing the instructions stored on the memory, wherein execution of the instructions programs the at least one processor to perform operations comprising:
electronically monitoring a telephonic interaction, the interaction comprising at least two different channels;
while monitoring the telephonic interaction, assigning one of a plurality of context-based speech recognition models to at least one of the two different channels wherein the one of the plurality of context-based speech recognition models is trained to recognize protected information and the plurality of context-based speech recognition models are organized in a multi-level hierarchical structure;
for at least one of the two different channels, swapping the assigned context-based speech recognition model with an alternate context-based speech recognition model while still monitoring the telephonic interaction, and wherein the speech from one of the other of the two different channels comprises a question, and wherein the alternate context-based speech recognition model comprises a fixed set of possible answers for the question; and
as the text is being transcribed, transcribing the monitored telephonic interaction from speech to text based on the assigned model such that the protected information is obfuscated.

### Claim 11 (depends on 10)

The system of claim 10, further comprising a data storage module for storing the transcribed text.

### Claim 12 (depends on 10)

The system of claim 10, further comprising a user interface for presenting the transcribed text to a user while the user participates in the telephonic interaction.

### Claim 13 (depends on 10)

The system of claim 10, further comprising a model storage module for storing the plurality of context-based speech recognition models.

### Claim 14 (depends on 10)

The system of claim 10, wherein one of the two different channels represents a service consumer and the other one of the two different channels represents a service provider.

### Claim 15 (depends on 10)

The system of claim 10, wherein the protected information is subject to legal protections under one or more statutes.

### Claim 16 (depends on 15)

The system of claim 15, wherein the one or more statutes comprise the General Data Protection Regulation (“GDPR”), the Health Insurance Portability and Accountability Act (“HIPAA”), or the Family Educational Rights and Privacy Act (“FERPA”).

### Claim 17 (depends on 10)

The system of claim 10, wherein a subset of the plurality of context-based speech recognition models include role-specific variants such that the assignment of one of the plurality of context-based speech recognition models to one of the channels is based in part on a transaction role of one of the channels.

### Claim 18 (depends on 10)

The system of claim 10, wherein the protected information comprises financial account data.

### Claim 19 (depends on 10)

The system of claim 10, wherein the protected information comprises healthcare data.

### Claim 20 (depends on 10)

The system of claim 10, wherein the alternate context-based speech recognition model assigned to at least one of the two different channels is based at least in part on speech from one of the other of the at least two different channels.