Related Experiment Video
Updated: Feb 7, 2026
03:31
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
1.1K
Improving competing voices segregation for hearing impaired listeners using a low-latency deep neural network
Lars Bramsløw1, Gaurav Naithani2, Atefeh Hafez1
1Eriksholm Research Centre, Oticon A/S, Rørtangvej 20, DK-3070 Snekkersten, Denmark.
The Journal of the Acoustical Society of America
|August 5, 2018
Summary
Deep neural networks (DNNs) improve speech intelligibility for hearing aid users in challenging multi-talker environments. DNNs enhance sound stream segregation, significantly boosting comprehension, especially when users can select their preferred voice.
Area of Science:
- Audiology
- Signal Processing
- Artificial Intelligence
Background:
- Hearing aid users struggle with speech segregation in noisy environments, particularly with multiple competing speakers.
- Distinguishing and focusing on a single voice among several is significantly harder for individuals with hearing impairments compared to those with normal hearing.
Purpose of the Study:
- To evaluate the effectiveness of a deep neural network (DNN) based algorithm for enhancing speech stream segregation in hearing-impaired listeners.
- To quantify the improvement in speech intelligibility provided by DNN-based sound separation in complex listening scenarios.
Main Methods:
- A DNN algorithm was employed to separate a mixture of two competing voices into individual streams.
- These separated streams were presented binaurally to hearing-impaired participants, and their word recognition scores were assessed.
- Performance was compared between the unseparated mixture and the DNN-separated streams, with and without target selection.
Main Results:
- DNN-based stream segregation yielded a 13% improvement in word score when participants attended to both voices.
- A significantly larger benefit of 37% was observed when participants selected a single target stream, mimicking traditional hearing aid scenarios.
- Results correlated well with objective metrics, indicating the algorithm's efficacy.
Conclusions:
- Deep neural networks show substantial potential for enhancing stream segregation and speech intelligibility for hearing aid users in challenging multi-talker situations.
- The DNN algorithm effectively improves comprehension even when users need to attend to two equally important targets without pre-selection.
- User-controlled target selection via remote further amplifies the benefits, suggesting a promising avenue for advanced hearing aid technology.
Related Concept Videos
01:31Hearing
57.3K
When we hear a sound, our nervous system is detecting sound waves—pressure waves of mechanical energy traveling through a medium. The frequency of the wave is perceived as pitch, while the amplitude is perceived as loudness.
57.3K
01:49Law of Segregation
78.2K
When crossing pea plants, Mendel noticed that one of the parental traits would sometimes disappear in the first generation of offspring, called the F1 generation, and could reappear in the next generation (F2). He concluded that one of the traits must be dominant over the other, thereby causing masking of one trait in the F1 generation. When he crossed the F1 plants, he found that 75% of the offspring in the F2 generation had the dominant phenotype, while 25% had the recessive phenotype.
78.2K
01:16Psychosexual Stages of Personality: Latency
1.3K
Following the phallic stage in Freud's theory of psychosexual development, children enter a phase called the latency period, which lasts from approximately six to twelve years of age. Unlike earlier stages, where sexual impulses played a central role, Freud believed these impulses are repressed during the latency period, becoming part of the unconscious. This stage is often described as a time of psychological calm after the turbulence of the phallic stage.
The latency period is not...
The latency period is not...
1.3K
01:16Segregation in Fresh Concrete
591
Segregation in fresh concrete is a phenomenon where the components of the concrete mix separate, leading to uneven distribution and compromised structural integrity. This separation typically occurs when concrete is subjected to excessive horizontal movement within forms, or when it is dropped from considerable heights or forced through narrow, winding paths. As a result, heavier coarse aggregate particles settle at the bottom, while lighter, finer materials such as cement and water rise to the...
591
02:26Protein Networks
4.6K
An organism can have thousands of different proteins, and these proteins must cooperate to ensure the health of an organism. Proteins bind to other proteins and form complexes to carry out their functions. Many proteins interact with multiple other proteins creating a complex network of protein interactions.
These interactions can be represented through maps depicting protein-protein interaction networks, represented as nodes and edges. Nodes are circles that are representative of a protein,...
These interactions can be represented through maps depicting protein-protein interaction networks, represented as nodes and edges. Nodes are circles that are representative of a protein,...
4.6K
01:12Trial and Error and Algorithm
424
A problem-solving strategy is a plan of action used to find a solution. Different strategies have distinct action plans. Trial and error involves trying different solutions until one works. For instance, to fix a broken printer, you might check ink levels, ensure the paper tray isn't jammed, and verify the printer's connection to your laptop. This method can be time-consuming but is commonly used. Thomas Edison, for example, used trial and error to find a suitable filament for the light...
424