Hybrid Expansive Frequency Compression for Speech Clarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Hearing aids have limited ability to provide sufficient amplification for high-frequency hearing loss, leading to difficulties in perceiving speech sounds like fricatives and affricates, which causes confusion between similar sounds, especially in noisy environments.
Innovation Solution
A method of audio signal processing using Hybrid Expansive Frequency Compression (hEFC) in digital signal processors, which detects high-frequency energy in speech sounds, classifies them into sound classes, and applies input-dependent frequency remapping to generate output frequencies that are either compressive and then expansive or expansive and then compressive, reducing sound frequency and enhancing speech perception.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If frequency lowering is applied to help individuals hear speech sounds, then audibility of high-frequency speech sounds is improved, but confusion between similar speech sounds increases
Solution Approach 1:
The patent segments the frequency lowering process into two distinct stages: first compressive frequency lowering to bring high-frequency speech sounds into the audible range, then expansive frequency spreading to separate similar speech sounds in the frequency domain. This segmentation resolves the contradiction by handling audibility and discrimination as separate processing steps rather than a single operation.
Solution Approach 2:
The patent applies dynamic, sound-class-specific processing where the frequency remapping function adapts based on the detected speech sound class. Different speech sounds undergo different frequency transformations, allowing the system to optimize both audibility and discrimination dynamically rather than using a static frequency lowering approach.
2Reliability
If high-frequency amplification is increased to overcome hearing loss, then audibility in high-frequency regions is improved, but distortion and discomfort increase
Solution Approach 1:
The patent introduces low-frequency regions as an intermediary domain. High-frequency speech sounds are first transformed into low-frequency regions through compressive frequency lowering, avoiding the need for direct high-frequency amplification that causes distortion and discomfort. The low-frequency region serves as a safe intermediate space for processing.
Solution Approach 2:
The patent changes the frequency parameter of speech sounds through compressive frequency lowering, transforming high-frequency sounds into lower frequency regions where they can be heard without the distortion and discomfort associated with high-frequency amplification. This parameter transformation resolves the contradiction by operating in a different frequency domain.
3Reliability
If compressive frequency lowering is applied to improve speech audibility, then high-frequency speech sounds become audible, but perceptual distinctiveness between similar sounds decreases
Solution Approach 1:
The patent applies dynamic frequency remapping where the expansion factor varies based on the detected speech sound class. Sounds that require greater separation undergo larger expansive transformations, while others receive smaller adjustments. This dynamic approach maintains perceptual distinctiveness while ensuring audibility.
Solution Approach 2:
The patent applies different frequency remapping characteristics to different speech sound classes. Each speech sound class receives a customized frequency transformation tailored to its specific perceptual requirements, ensuring that similar sounds are adequately separated while maintaining overall speech intelligibility.
Data Source
AI summary
A method of audio signal processing comprising Hybrid Expansive Frequency Compression (hEFC) via a digital signal processor, wherein the method includes: classifying an audio signal input, wherein the audio signal input includes frication high-frequency speech energy, into two or more speech sound classes followed by selecting a form of input-dependent frequency remapping function; and performing hEFC including, re-coding of one or more input frequencies of the speech sound via the input-dependent frequency remapping function to generate an audio output signal, wherein the output signal is a representation of the audio signal input having a lower sound frequency.