Process Dataset

Python Modules

recipes.cad2.task2.process_dataset.generate_train_scenes

Module to generate new train scenes

recipes.cad2.task2.process_dataset.process_zenodo_download

Module take the data downloaded from Zenodo and rearrange the data for the challenge.

Configuration Files

config.yaml

 1# Zenodo download path: path to the zenodo download folder.
 2# root: root path of the dataset. This path will contain the audio and metadata folders
 3path:
 4  zenodo_download_path: ??
 5  root: ??
 6  metadata_dir: ${path.root}/metadata
 7  scene_metadata_file: ${path.metadata_dir}/scenes.train2.json
 8  gain_metadata: ${path.metadata_dir}/gains_meta.json
 9  music_dir: ${path.root}/audio
10  music_metadata: ${path.metadata_dir}/music.train.json
11
12seed_scene_generation: Cad2Task2_2024
13gains_per_track: 4
14listener_per_scene: 2
15
16# hydra config
17hydra:
18  job:
19    chdir: True