# US 12,014,740 B2 · Systems and methods for contactless authentication using voice recognition

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

- Number: US 12,014,740 B2
- Document ID: US12014740B2
- Status: Granted
- Kind: B2
- Application number: 16/820,010
- Filed: 2020-03-16
- Published: 2021-09-16
- Granted: 2024-06-18
- Earliest claimed: 2019-01-08
- Assignees: Fidelity Information Services, LLC
- Applicants: Fidelity Information Services, LLC
- Inventors: Raghavendra Pratap Singh; Vijayendra Virendra Mishra
- Web page: https://patentgrep.com/patent/US12014740B2

## CPC

- G10L 17/04: Training, enrolment or model building
- G06F 21/32: using biometric data, e.g. fingerprints, iris scans or voiceprints
- G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/0475: Generative networks
- G06N 3/082: modifying the architecture, e.g. adding, deleting or silencing nodes or connections
- G06N 3/09: Supervised learning
- G06N 3/094: Adversarial learning
- G06N 3/0985: Hyperparameter optimisation; Meta-learning; Learning-to-learn
- G10L 17/18: Artificial neural networks; Connectionist approaches
- G10L 17/22: Interactive procedures; Man-machine interfaces
- H04L 63/0861: {using biometrical features, e.g. fingerprint, retina-scan (cryptographic mechanisms or cryptographic arrangements for entity authentication using biological data H04L9/3231)}
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06N 20/00: Machine learning
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

Systems and methods for contactless authorization using voice recognition is disclosed. The system may include one or more memory units storing instructions and one or more processors configured to execute the instructions to perform operations. The operations may include receiving user data comprising a user identifier, an audio data having a first data format, and a client device identifier. The operations may include generating a processed audio data based on the received audio data. The processed audio data may have a second data format. The operations may include transmitting, to a speech module, the processed audio data. The operations may include receiving from the speech module, a voice match result. In some embodiments, the operations include authenticating a user based on the voice match result and transmitting, to a client device associated with the client device identifier, a client notification comprising a result of the authentication.

## Claims

### Claim 1

An authentication system comprising:
at least one memory storing instructions; and
one or more processors configured to execute the instructions to perform operations comprising:
receiving a speech authentication model request from a user device of a user, the speech authentication model request comprising training data including audio data of the user having an original audio data format with an identifier and a training command to train a speech authentication model for deployment at the user device;
receiving instructions for the training command, wherein the instructions comprise generate an ephemeral container instance to execute the training command;
processing the audio data of the user by converting the audio data from the original audio data format into a different audio data format;
based on the training command, training a speech authentication model, using the training data including the converted audio data comprising speech of the user and audio data labeled as containing a passphrase, to determine a match between a received voice signal and the processed audio data of the user when the speech authentication model is deployed at the user device, the passphrase being authentication data for contactless authentication by the user device;
determining an accuracy of the trained speech authentication model; and
transmitting the trained speech authentication model to the user device for deployment at the user device.

### Claim 2 (depends on 1)

The authentication system of claim 1, wherein the speech authentication model is a stored speech authentication model previously trained to match voice signals.

### Claim 3 (depends on 1)

The authentication system of claim 1, the operations further comprising generating the speech authentication model prior to training the trained speech model.

### Claim 4 (depends on 1)

The authentication system of claim 1, the operations further comprising storing the speech authentication model.

### Claim 5 (depends on 1)

The authentication system of claim 1, wherein the speech authentication model request is received at a storage system component of the authentication system, and the storage system component performs the processing of the audio data.

### Claim 6 (depends on 5)

The authentication system of claim 5, the operations further comprising transmitting an alert from the storage system component to an authentication module of the authentication system, the alert comprising the processed audio data.

### Claim 7 (depends on 6)

The authentication system of claim 6, the operations further comprising destroying processed audio data stored on the storage system component.

### Claim 8 (depends on 1)

The authentication system of claim 1, the operations further comprising:
receiving a matching request from the user device, the matching request comprising additional audio data;
processing the additional audio data by converting the additional audio data into the different audio data format;
generating a match result based on the processed additional audio data;
updating the speech model based on the match result; and
transmitting the match result to the user device.

### Claim 9 (depends on 8)

The authentication system of claim 8, wherein transmitting the match result comprises transmitting the updated speech model.

### Claim 10 (depends on 1)

The authentication system of claim 1, wherein processing the audio data is performed by an ephemeral container instance of the authentication system.

### Claim 11 (depends on 1)

The authentication system of claim 1, wherein the speech authentication model comprises at least one of a recurrent neural network model, a hidden Markov model, a discriminative learning model, a Bayesian learning model, a structured sequence learning model, or an adaptive learning model.

### Claim 12 (depends on 1)

The authentication system of claim 1, wherein the speech authentication model is a previously trained speech model associated with user data.

### Claim 13 (depends on 1)

The authentication system of claim 1, wherein the authentication system comprises at least one scalable cloud service configured to generate and terminate container instances.

### Claim 14 (depends on 1)

The authentication system of claim 1, wherein training the model comprises using training data comprising the processed audio data and audio data associated with a plurality of individuals.

### Claim 15 (depends on 1)

The authentication system of claim 1, wherein training the model includes optimizing a model parameter and a hyperparameter.

### Claim 16 (depends on 1)

The authentication system of claim 1, the operations further comprising storing the processed audio data as reference data associated with a user.

### Claim 17 (depends on 1)

The authentication system of claim 1, wherein:
the operations include transmitting, from an authentication module of the authentication system to a speech module of the authentication system, instructions to train the speech authentication model, and
training the speech authentication model is performed by the speech module.

### Claim 18 (depends on 1)

The authentication system of claim 1, wherein the speech authentication model comprises an algorithm to recognize speech of the user.

### Claim 19

A computer-implemented method comprising:
receiving a speech authentication model request from a user device of a user, the speech authentication model request comprising training data including audio data of the user having an original audio data format with an identifier and a training command to train a speech authentication model for deployment at the user device;
receiving instructions for the training command, wherein the instructions comprise generate an ephemeral container instance to execute the training command;
processing the audio data of the user by converting the audio data from the original audio data format into a different audio data format;
based on the training command, training a speech authentication model, using the training data including the converted audio data comprising speech of the user and audio data labeled as containing a passphrase, to determine a match between a received voice signal and the processed audio data of the user when the speech authentication model is deployed at the user device, the passphrase being authentication data for contactless authentication by the user device;
determining an accuracy of the trained speech authentication model; and
transmitting the trained speech authentication model to the user device for deployment at the user device.

### Claim 20

An authentication system comprising:
at least one memory storing instructions; and
one or more processors configured to execute the instructions to perform operations comprising:
receiving a speech authentication model request from a user device of a user, the speech authentication model request comprising training data including audio data of the user having an original audio data format with an identifier and a training command to train a speech authentication model for deployment at the user device;
receiving instructions for the training command, wherein the instructions comprise generate an ephemeral container instance to execute the training command;
processing the audio data of the user by converting the audio data from the original audio data format into a different audio data format;
retrieving a stored speech model associated with the user data;
based on the training command, training the stored speech authentication model, using the training data including the converted audio data comprising speech of the user and audio data labeled as containing a passphrase, the passphrase being authentication data for contactless authentication by the user device, to determine a match between a received voice signal and the processed audio data of the user when the speech authentication model is deployed at the user device, the training further comprising using training data comprising the processed audio data and audio data associated with a plurality of individuals;
storing the trained speech model and the processed audio data;
determining an accuracy of the trained speech authentication model; and
transmitting the trained speech authentication model to the user device for deployment at the user device.