DCASE2022-Task1-Rion

This is a repostiory inteded the RION submission for DCASE2022 Task 1: Low-Complexity Acoustic Scene Classification. Technical report is here.

Folder Structure

├── TAU-urban-acoustic-scenes-2022-mobile-development      <- dataset 
│   ├── audio                                              <- audio data
│   ├── evaluation_setup        　　　　　                  <- file path and label    
└── DCASE2022-Task1-Rion                                   
    ├── dockerfile                                         <- Docker file
    ├── NeSsi                                              <- Model size calculation　https://github.com/AlbertoAncilotto/NeSsi
    │── model                                              <- model
    │── reuse                                              <- input feature as numpy file
    │── main.ipynb                                         <- main code
    │── estimate_devices_freq.ipynb                        <- for spectrum modulation
    │── config_submit1.py                                  <- config file
    │── pcgrad.py                                          <- multi-task learning system https://github.com/WeiChengTseng/Pytorch-PCGrad
    │── README.md
    └── .gitignore                                         <- Specifies intentionally untracked files to ignore