MSc thesis project proposal

Small-footprint Embedded Real-Time Speech Enhancement for Cochlea Implant

Cochlear implants (CI) [1, 2] are miniaturized biomedical devices that can help deaf people perceive sound or help hearing loss patients understand speech better. The CI has an in-vitro module attached behind the ear and an in-vivo implant surgically placed under the skin. The quality of CI output signals degrades in noisy environments and relies on Speech enhancement (SE) systems to enhance its performance. Neural network-based SEs [3] achieve state-of-the-art performance but are expensive to deploy on CI with a limited power budget. In this work, you will build a small-footprint NN-based SE system that can run on embedded devices [4, 5, 6] in real-time.


Background Materials

  1. Hochmair, I., Nopp, P., Jolly, C., Schmidt, M., Schösser, H., Garnham, C., & Anderson, I. (2006). MED-EL Cochlear implants: state of the art and a glimpse into the future. Trends in amplification, 10(4), 201–219.

  2. Joseph Tierny, Marc A. Zissman, and Donald K. Eddington. 1994. Digital signal processing applications in cochlear-implant research. <i>Lincoln Lab. J.</i> 7, 1 (Spring/Summer 1994), 31–62.

  3. Nengheng Zheng, et al., “A noise-robust signal processing strategy for cochlear implants using neural networks”, ICASSP 2021.

  4. Portenta H7 — Arduino Official Store

  5. MiniZed | Avnet Boards


  • Survey the CI market to understand the performance requirements of SE and its latency & power constraints.
  • Design a small-footprint neural network-based SE algorithm or optimize an existing SE system.
  • Implement the SE in an embedded platform (μC or FPGA) with real-time operation capability. (CUDA/High-Level Synthesis/SystemVerilog)


  • Experience with Python and PyTorch.
  • Experience in using μC is a plus.
  • Digital design with Verilog or Xilinx HLS (optional).


dr. Chang Gao

Electronic Circuits and Architectures Group

Department of Microelectronics

Last modified: 2022-10-10