# US 2018/0322882 A1 · Method and system for using conversational biometrics and speaker identification/verification to filter voice streams

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

- Number: US 2018/0322882 A1
- Document ID: US20180322882A1
- Status: Published application
- Kind: A1
- Application number: 16/031,170
- Filed: 2018-07-10
- Published: 2018-11-08
- Earliest claimed: 2008-10-06
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Peeyush JAISWAL; Naveen NARAYAN
- Web page: https://patentgrep.com/patent/US20180322882A1

## CPC

- G10L 17/00: Speaker identification or verification techniques
- G10L 17/02: Preprocessing operations, e.g. segment selection; Pattern representation or modelling, e.g. based on linear discriminant analysis [LDA] or principal components; Feature selection or extraction
- G10L 17/08: Use of distortion metrics or a particular distance between probe pattern and reference templates
- G10L 17/14: Use of phonemic categorisation or speech recognition prior to speaker recognition or verification
- G10L 21/0272: Voice signal separating
- H04M 3/42221: {Conversation recording systems (at the subscriber's set H04M1/656)}
- H04M 3/4936: {Speech interaction details (speech recognition per se G10L15/00)}
- H04M 3/5166: {in combination with interactive voice response systems or voice portals, e.g. as front-ends}
- H04M 3/5175: {Call or contact centers supervision arrangements}
- H04M 2201/41: using speaker recognition
- 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 method and system for using conversational biometrics and speaker identification and/or verification to filter voice streams during mixed mode communication. The method includes receiving an audio stream of a communication between participants. Additionally, the method includes filtering the audio stream of the communication into separate audio streams, one for each of the participants. Each of the separate audio streams contains portions of the communication attributable to a respective participant. Furthermore, the method includes outputting the separate audio streams to a storage system.

## Claims

### Claim 1

A method implemented in a computing system, the method comprising:
extracting a plurality of audio streams from a communication corresponding to a plurality of participants;
matching portions of the communication in the plurality of audio streams to voice prints which correspond to identified participants within the communication; and
adapting a speaker model of the voice prints after successfully matching the portions of the communication in the plurality of audio streams to the voice prints, wherein adapting the speaker model includes capturing long-term voice changes of the identified participants in the voice prints used for the matching.

### Claim 2 (depends on 1)

The method of claim 1, further comprising performing a verification process for at least one of the plurality of participants.

### Claim 3 (depends on 1)

The method of claim 1, wherein the voice prints include seed phrases received for each of the plurality of participants.

### Claim 4 (depends on 1)

The method of claim 1, wherein:
providers of the voice prints are each associated with a role; and
each of the voice prints is stored in one of a plurality of discrete databases based on the role of the associated provider.

### Claim 5 (depends on 1)

The method of claim 1, wherein the matching comprises:
matching the one or more of the portions of the communication to the voice prints using at least one of conversational biometrics, speaker identification and speaker verification; and
assigning the one or more portions of the communication attributable to each of the plurality of participants to the separate audio streams corresponding to each of the plurality of participants, wherein the plurality of audio streams contains portions of the communication corresponding to the plurality of participants.

### Claim 6 (depends on 1)

The method of claim 1, further comprising verifying at least one of the plurality of participants by authenticating a match between a voice sample of the at least one of the plurality of participants and a previously stored voice print of the at least one of the plurality of participants.

### Claim 7 (depends on 1)

The method of claim 1, further comprising receiving the plurality of audio streams from a sensor.

### Claim 8 (depends on 1)

The method of claim 1, further comprising converting the communication from speech to text.

### Claim 9 (depends on 1)

The method of claim 1, wherein the communication is a call center communication and the plurality of participants includes at least two of: a caller, an agent and an interactive voice response (IVR) system.

### Claim 10 (depends on 9)

The method of claim 9, further comprising collecting seed phrases for the agent and the IVR system prior to receiving the plurality of audio streams.

### Claim 11 (depends on 1)

The method of claim 1, wherein the communication is a medical professional communication and the plurality of participants include at least a medical professional and a patient.

### Claim 12 (depends on 1)

The method of claim 1, wherein the extracting comprises using the voice prints for speaker identification of all but one of the plurality of participants and a process of elimination for the one of the plurality of participants.

### Claim 13 (depends on 1)

The method of claim 1, wherein the extracting the plurality of audio streams comprises one of: performed in real-time and performed in a batch.

### Claim 14 (depends on 1)

The method of claim 1, wherein the extracting comprises determining an origin of a voice of at least one of the plurality of participants.

### Claim 15 (depends on 1)

The method of claim 1, further comprising filtering the plurality of audio streams into separate audio streams corresponding, respectively, to each of the plurality of participants in the communication, wherein each of the separate audio streams contains portions of the communication attributable to the corresponding one of the plurality of participants.

### Claim 16 (depends on 15)

The method of claim 15, further comprising identifying the plurality of participants by matching one or more of the portions of the communication in each of the separate audio streams to the voice prints.

### Claim 17 (depends on 16)

The method of claim 16, wherein the matching comprises comparing the separate audio streams to only the plurality of the voice prints corresponding to the identified participants within the communication.

### Claim 18 (depends on 1)

The method of claim 1, further comprising processing the voice prints by utilizing any one of: frequency estimation, hidden Markov models, Gaussian mixture models, pattern matching methodologies, neural networks, matrix representation, and decision trees, on the voice prints.

### Claim 19 (depends on 1)

The method of claim 1, further comprising utilizing text-dependent recognition, wherein a text of the voice print during an enrollment phase is identical to the text of the voice print during a verification phase.

### Claim 20 (depends on 1)

The method of claim 1, further comprising utilizing text-independent recognition, wherein a text of the voice print during an enrollment phase is different from the text of the voice print during a verification phase.
