Multi-Microphone Speech Augmentation Across Near- and Far-Field Systems

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Mismatches in speech signals processed by different microphones in multi-microphone systems lead to significant performance degradations in automated clinical documentation systems, particularly when transitioning from near-field to far-field microphone systems.

Innovation Solution

A data augmentation process that uses acoustic relative transfer functions to augment speech signals in real-time, accounting for reverberation and noise characteristics, enabling better alignment of speech data across different microphone systems.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If data augmentation is used to improve robustness to noise and reverberation, then speech processing accuracy is improved, but mismatch between speech signals from different microphones causes performance degradation

Engineering Contradiction:

Improvespeech processing accuracyVSAvoidcompatibility across microphone systems

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The system applies acoustic relative transfer functions to transform speech signals by changing acoustic parameters such as reverberation characteristics and noise profiles. This allows training data from one microphone system to be adapted to match the acoustic characteristics of another microphone system, resolving the mismatch issue while maintaining robustness improvements

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent introduces acoustic relative transfer functions as an intermediary transformation layer between speech signals from different microphone systems. This intermediary process enables compatibility by mapping signals from a first microphone system to match the characteristics of a second microphone system, allowing data augmentation to work effectively across heterogeneous hardware

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If speech signals are processed to account for reverberation and noise characteristics, then robustness is enhanced, but processing complexity increases

Engineering Contradiction:

Improverobustness to noise and reverberationVSAvoidsignal processing complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system performs preliminary processing by pre-computing and storing acoustic relative transfer functions that capture reverberation and noise characteristics. During runtime, these pre-computed functions are applied to speech signals, avoiding the need for complex real-time analysis of acoustic environments and reducing processing complexity while maintaining robustness

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS12456456B2Data augmentation system and method for multi-microphone systems

Publication Date: 2025.10.28 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

A method, computer program product, and computing system for obtaining one or more speech signals from a first device, thus defining one or more first device speech signals. One or more speech signals may be obtained from a second device, thus defining one or more second device speech signals. An acoustic relative transfer function may be selected from a plurality of acoustic relative transfer functions based upon, at least in part, the one or more first device speech signals and the one or more second device speech signals. The one or more second device speech signals may be augmented, at run-time, based upon, at least in part, the acoustic relative transfer function.