Baseline of DCASE 2020 task 4

Couple Learning for SED

• This repository provides the data and source code for sound event detection (SED) task.

• The improvement of the Couple Learning method is verified on the basis of the dcase20-task4 baseline.

• Information about the dcase20-task4 please visit github.

• Information about Couple Learning please visit paper: Couple Learning: Mean Teacher method with pseudo-labels improves semi-supervised deep learning results.

Couple Learning model

More info in the PLG-MT_run folder.

Reproducing the results

See PLG-MT_run folder.

Dependencies

Python >= 3.6, pytorch >= 1.0, cudatoolkit>=9.0, pandas >= 0.24.1, scipy >= 1.2.1, pysoundfile >= 0.10.2, scaper >= 1.3.5, librosa >= 0.6.3, youtube-dl >= 2019.4.30, tqdm >= 4.31.1, ffmpeg >= 4.1, dcase_util >= 0.2.5, sed-eval >= 0.2.1, psds-eval >= 0.1.0, desed >= 1.3.0

A simplified installation procedure example is provided below for python 3.6 based Anconda distribution for Linux based system:

1. install Ananconda
2. launch conda_create_environment.sh (recommended line by line)

Dataset

All the scripts to get the data (soundbank, generated, separated) are in the scripts folder and they use python files from data_generation folder.

Scripts to generate the dataset

In the scripts/ folder, you can find the different steps to:

• Download recorded data and synthetic material.
• Generate synthetic soundscapes
• Reverberate synthetic data (Not used in the baseline)
• Separate sources of recorded and synthetic mixtures

It is likely that you'll have download issues with the real recordings. At the end of the download, please send a mail with the TSV files created in the missing_files directory.

However, if none of the audio files have been downloaded, it is probably due to an internet, proxy problem. See Desed repo or Desed_website for more info.

Base dataset

The dataset for sound event detection of DCASE2020 task 4 is composed of:

• Train:
  • *weak (DESED, recorded, 1 578 files)
  • *unlabel_in_domain (DESED, recorded, 14 412 files)
  • synthetic soundbank (DESED, synthetic, 2060 background (SINS only) + 1006 foreground files)
• *Validation (DESED, recorded, 1 168 files):
  • test2018 (288 files)
  • eval2018 (880 files)

Baselines dataset

SED baseline

• Train:
  • weak
  • unlabel_in_domain
  • synthetic20/soundscapes (separated in train/valid-80%/20%)
• Validation:
  • validation