This is our final year design project, which focuses on improving the performance of hearing aids using neural beamforming and source separation techniques.
The goal of this project is to enhance the listening experience of hearing aid users by localizing speakers and separating out background noise. Our approach uses a neural beamformer to locate the direction of the speaker, then applly source separation to isolate the speaker's voice from surrounding sounds. This results in a clearer, more intelligible audio signal for the user.
- Rishi Kakade
- Aditya Srinivasan
- Yulia Kluev
- Ojorumi Aneke
- Kiersten Overton
- Kunal Chandan
Our beamformer was Cone of Silence and the source separation was done using Conv-TasNet