Voice Processing Device Integrating Frequency Response for Speaker Verification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Speaker verification accuracy decreases when different input devices are used for registration and verification due to varying frequency sensitivity, leading to potential verification failures.

Innovation Solution

A voice processing device integrates voice data with the frequency response of the input device to extract a speaker feature, using a deep neural network to concatenate acoustic and device features, enabling accurate verification regardless of the input device used.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If different input devices are used for registration and verification, then device versatility is improved, but speaker verification accuracy deteriorates

Engineering Contradiction:

Improveinput device compatibilityVSAvoidspeaker verification accuracy

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent introduces frequency response information as an intermediary element that mediates between the voice data and the verification process. By incorporating the frequency response characteristics of the input device, the system compensates for device-specific variations, enabling accurate speaker verification across different input devices without requiring the same device for registration and verification

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent changes the parameter set used for verification by adding frequency response parameters to the traditional voice feature parameters. This parameter expansion allows the verification system to account for and compensate for differences in input device characteristics, thereby maintaining verification accuracy across diverse devices

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If frequency response integration is added to the verification process, then speaker verification accuracy is improved, but processing complexity increases

Engineering Contradiction:

Improvespeaker verification accuracyVSAvoidprocessing complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by obtaining and storing the frequency response information of the input device before the verification process. This pre-acquired information is then integrated with the voice data during verification, avoiding the need for complex real-time calculations and reducing processing complexity while maintaining accuracy

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20230326465A1Voice processing device, voice processing method, recording medium, and voice authentication system

Publication Date: 2023.10.12 NEC CORP

AI summary

The present disclosure implements speaker verification with high accuracy regardless of input devices. An integration unit (110) integrates voice data inputted using an input device, and the frequency characteristic of the input device, and a feature extraction unit (120) extracts, from an integrated feature obtained by integrated the voice data and the frequency characteristic, a speaker feature for verifying the speaker of voice.