Related Experiment Video
Updated: May 11, 2026
Enhanced Cochlear Coverage and Hearing Preservation in High-Frequency Hearing Loss via Electric Acoustic Stimulation with Longer Electrode
Published on: October 11, 2024
Effect of spectral change enhancement for the hearing impaired using parameter values selected with a genetic
Jing Chen1, Thomas Baer, Brian C J Moore
1Department of Machine Intelligence, Speech and Hearing Research Center and Key Laboratory of Machine Perception (Ministry of Education), Peking University, Beijing 100871, People's Republic of China. janechenjing@gmail.com
Personalizing speech enhancement algorithms significantly improved hearing-impaired individuals' ability to understand speech in noise. Tailoring processing parameters using a genetic algorithm yielded better results than fixed settings.
Area of Science:
- Audiology
- Speech processing
- Signal processing
Background:
- Hearing loss often impairs speech intelligibility in noisy environments.
- Previous spectral enhancement algorithms showed mixed results, improving intelligibility in steady speech-spectrum noise (SSN) but not in two-talker speech (TTS).
- Significant individual differences in response to speech enhancement were observed.
Purpose of the Study:
- To investigate if individually tailoring speech enhancement algorithm parameters using a genetic algorithm improves speech intelligibility for hearing-impaired listeners.
- To compare the effectiveness of personalized versus fixed enhancement parameters across different noise conditions (SSN and TTS) and signal-to-masker ratios (SMRs).
Main Methods:
- A genetic algorithm was employed to select optimal enhancement parameters based on individual listener preferences for speech clarity.
- Speech intelligibility was measured for both unprocessed and processed speech stimuli.
- Stimuli included speech in steady speech-spectrum noise (SSN) and two-talker speech (TTS) maskers at two signal-to-masker ratios (SMRs).
Main Results:
- Parameter values selected by the genetic algorithm varied considerably among subjects.
- Speech intelligibility in SSN at the lower SMR improved by approximately 14 percentage points with personalized processing.
- The overall improvement with personalized parameters was significantly greater than that achieved with fixed parameters in prior studies.
Conclusions:
- Individualized parameter selection for speech enhancement algorithms, guided by a genetic algorithm, is beneficial for improving speech intelligibility in hearing-impaired listeners.
- Personalization effectively enhances speech understanding in specific noise types, particularly in challenging listening conditions (lower SMR).
- This approach offers a promising strategy for developing more effective hearing assistive technologies.
