Related Experiment Video
Updated: Jun 10, 2025
06:04
Systematic Hearing Performance Evaluation Process for Adolescents with Cochlear Implantation at Early Ages
Published on: March 24, 2023
343
Employing deep learning model to evaluate speech information in acoustic simulations of Cochlear implants
1Department of Otolaryngology, New York University Grossman School of Medicine, 550 First Avenue, New York, NY, 10016, USA.
Scientific Reports
|October 14, 2024
Summary
This study uses a deep learning model to assess cochlear implant (CI) speech simulations, finding it mimics human responses to parameter changes. This approach offers a faster, more cost-effective alternative to traditional human subject testing for auditory research.
Area of Science:
- Auditory Neuroscience
- Speech Processing
- Artificial Intelligence
Background:
- Acoustic vocoders simulate speech for cochlear implant (CI) research.
- Traditional intelligibility testing with human subjects is resource-intensive.
- A need exists for efficient and reliable methods to evaluate CI simulations.
Purpose of the Study:
- To investigate the intelligibility of CI simulations using a deep learning speech recognition model.
- To evaluate the model's response to variations in vocoder parameters.
- To establish an alternative to human subject testing for CI research.
Main Methods:
- Utilized an advanced deep learning speech recognition model.
- Evaluated vocoder-processed words and sentences with adjusted parameters (bands, frequency range, dynamic range).
- Simulated psychophysical temporal and intensity resolutions by manipulating envelope properties.
Main Results:
- The deep learning model demonstrated human-like responses to changes in vocoder parameters.
- Model performance correlated with alterations in simulated CI processing settings.
- Results aligned with existing human subject data, validating the model's approach.
Conclusions:
- Deep learning models can effectively assess the intelligibility of cochlear implant simulations.
- This AI-driven method offers significant time and cost savings over human testing.
- Speech recognition models show promise for advancing auditory research and development.
