Neural Network Voice Segment Detection in Noise

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for determining voice segments in input signals are inaccurate in noisy environments due to noise interference, affecting sound volume measurements.

Innovation Solution

A segment detecting device utilizing a neural network to estimate the level of a target signal within an input signal, comparing this level to a threshold to determine valid segments, and dynamically adjusting the threshold based on estimated levels and signal characteristics.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If sound volume of input signal is used for determining voice segment, then the determination process is simple, but the accuracy deteriorates under noise environment

Engineering Contradiction:

Improvedetermination process complexityVSAvoidvoice segment determination accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent introduces a neural network as an intermediary component between the input signal and the voice segment determination. The neural network processes the input signal to extract features and determine voice segments, acting as a mediator that separates the simple input from the complex determination task, thereby improving accuracy without significantly increasing overall system complexity

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces the simple sound volume measurement mechanism with a neural network-based processing system. Instead of directly using sound volume for determination, the system uses the neural network to analyze signal characteristics and estimate noise levels, substituting a mechanical measurement approach with an intelligent processing approach to overcome noise interference

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If neural network is used to estimate signal level, then the determination accuracy is improved, but the device complexity increases

Engineering Contradiction:

Improvevoice segment determination accuracyVSAvoidsystem structure complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by having the neural network trained in advance with large amounts of data before deployment. The training process is performed separately and beforehand, allowing the neural network to be pre-configured with optimal parameters and knowledge, so that during actual operation, only inference is needed, reducing real-time computational complexity

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The neural network performs self-service by automatically adapting to different noise environments and signal conditions. Once trained, the network autonomously estimates noise levels and determines voice segments without requiring manual intervention or complex external control mechanisms, making the system self-sufficient in handling varied conditions

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS12142262B2Segment detecting device, segment detecting method, and model generating method

Publication Date: 2024.11.12 PREFERRED NETWORKS INC

AI summary

A segment detecting device according to an embodiment includes at least one memory; and at least one processor. The at least one processor receives at least one of (i) an input signal including a first signal and a second signal or (ii) feature data representing one or a plurality of features of the input signal, estimates a level of the second signal by inputting the input signal or the feature data into a neural network, and determines a segment including the second signal in the input signal based on the level of the second signal.