Single-Microphone-Based Sound Source Localization for Mobile Robots in Reverberant Environments

• This is an online sound source localization method that uses a single microphone mounted on a mobile robot in reverberant environments.

• Specifically, we develop a lightweight neural network model with only 43k parameters to perform real-time distance estimation by extracting temporal information from reverberant signals. The estimated distances are then processed using an extended Kalman filter to achieve online sound source localization.

Problem setup and sound source localization framework

Structure of the distance estimation neural network

Real World Setup

Detailed Usage

Distance Estimations

# 1. Prepare simulation datasets by FRAM_RIR script (https://github.com/tencent-ailab/FRA-RIR)
python produce_RIR_1_room.py      # 1 room
# or
python produce_RIR_100_room.py    # 100 rooms 

# 2. Change the paths in the files under the foders of "chirp_signals", "configs"

# 3. Train the model with simulation dataset
python train_dis_simu.py

# 4. Test the model with simulation dataset
python test_dis_sim.py

Sound source localization with real world datasets

# 1. Train the simulation model under 100 rooms

# 2. Fine tune the model with real world datasets
python train_dis_real.py

# 3. Test the model with real world dataset
python test_dis_real.py

# 4. Estimate the sound source positions using the jupyter file "sound_source_localization.ipynb"

License

The source code and dataset are released under MIT license.

Citation

Please cite the paper if you use the codes or data for your research.

@INPROCEEDINGS{2025SSL,
  author={Wang, Jiang and Shi, Runwu and Yen, Benjamin and Kong, He and Nakadai, Kazuhiro},
  booktitle={2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)}, 
  title={Single-Microphone-Based Sound Source Localization for Mobile Robots in Reverberant Environments}, 
  year={2025},
  pages={6135-6140},
  doi={10.1109/IROS60139.2025.11246992}}