Cochlear Implant Transient Sound Classification and Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current cochlear implant systems face challenges in effectively distinguishing and processing transient sounds, which can lead to reduced speech intelligibility and discomfort due to background noise, as existing noise reduction algorithms often fail to accurately differentiate between transient noise and speech features.

Innovation Solution

A method is introduced to generate electrode stimulation signals by processing audio signals to detect and modify channel envelopes based on whether the transient is noise or speech, using a filter bank to reduce noise and enhance speech features, thereby improving hearing comfort and speech intelligibility.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Object-affected harmful factors

If noise reduction algorithms are applied to reduce transient noise, then listening comfort is improved, but speech intelligibility deteriorates because transient speech features are also attenuated

Engineering Contradiction:
Improvetransient noiseVSAvoidspeech intelligibility
Core Design Contradiction:
Object-affected harmful factorsVSLoss of information

Solution Approach 1:

The patent applies spectral shaping to selectively modify the frequency content of transient sounds. By analyzing the spectral characteristics and temporal envelope of transients, the system differentiates between noise transients (which are suppressed) and speech transients (which are preserved or enhanced), effectively 'color-coding' different transient types through frequency-specific gain adjustments

Inventive Principle:
Principle #32Color changes

Solution Approach 2:

The system dynamically changes multiple parameters including spectral centroid, bandwidth, rise time, and decay time to characterize and differentiate transient sounds. These parameter changes enable the algorithm to distinguish between transient noise and transient speech, allowing selective processing that reduces noise while preserving speech intelligibility

Inventive Principle:
Principle #35Parameter changes

2Loss of information

If transient enhancement is applied to improve speech features, then speech intelligibility is improved, but listening comfort deteriorates due to amplified transient noise

Engineering Contradiction:
Improvespeech intelligibilityVSAvoidtransient noise
Core Design Contradiction:
Loss of informationVSObject-affected harmful factors

Solution Approach 1:

The patent uses spectral shaping to selectively enhance or suppress different frequency regions based on the identified transient type. For transient speech, the system enhances the relevant frequency bands while leaving transient noise frequencies unaffected, achieving selective enhancement that improves speech intelligibility without amplifying noise

Inventive Principle:
Principle #32Color changes

Solution Approach 2:

The system applies different processing characteristics to different frequency bands and transient types. By analyzing local spectral and temporal features, the algorithm applies enhancement only where speech transients are detected, while applying suppression where noise transients are identified, creating locally optimized processing for each frequency region

Inventive Principle:
Principle #3Local quality

3Object-affected harmful factors

If existing noise reduction algorithms process all transients uniformly, then transient noise is reduced, but the system fails to distinguish between transient noise and transient speech

Engineering Contradiction:
Improvetransient noiseVSAvoidtransient classification accuracy
Core Design Contradiction:
Object-affected harmful factorsVSMeasurement precision

Solution Approach 1:

The patent employs multiple acoustic parameters including spectral centroid, bandwidth, rise time, decay time, and zero-crossing rate to characterize transient sounds. By monitoring changes in these parameters, the system accurately distinguishes between transient noise and transient speech, achieving precise classification that enables selective processing

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The system analyzes transients in both the frequency domain (spectral characteristics) and time domain (temporal envelope, rise/decay times). This multi-dimensional analysis approach adds discriminatory power beyond simple amplitude thresholding, enabling accurate differentiation between noise and speech transients based on their distinct spectral-temporal signatures

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Data Source

PatentEP3079760B1Automatic selection of reduction or enhancement of transient sounds
Publication Date: 2020.08.12 MED EL ELEKTROMEDIZINISCHE GERAETE GMBH
  • EP3079760B1 patent drawingFigure 1
  • EP3079760B1 patent drawingFigure 2
  • EP3079760B1 patent drawingFigure 3(a)~3(b)

AI summary

A system and method of generating electrode stimulation signals for electrode contacts in an electrode array associated with a hearing implant is presented. An input audio signal is processed to generate a plurality of band pass channel signals each representing an associated band of audio frequencies. A stationary noise reduction is applied so as to provide a stationary noise reduced channel envelope from each channel signal. A transient in one or more of the channel envelopes is detected. The channel envelopes are modified as a function of whether the transient is transient noise or transient speech, so as to form transient modified envelope. The transient modified envelopes are used to generate electrode stimulation signals to the electrode contacts.