config/augmentations/default.yaml

# @package __global__

audio_effects:
  speed:
    sample_rate: ${sample_rate}
    speed_range: [0.8, 1.2]
  updownresample:
    sample_rate: ${sample_rate}
    intermediate_freq: 32000
  echo:
    sample_rate: ${sample_rate}
    volume_range: [0.1, 0.5]
    duration_range: [0.1, 0.5]
  random_noise:
    noise_std: 0.001
  pink_noise:
    noise_std: 0.01
  lowpass_filter:
    sample_rate: ${sample_rate}
    cutoff_freq: 5000
  highpass_filter:
    cutoff_freq: 500
    sample_rate: ${sample_rate}
  bandpass_filter:
    cutoff_freq_low: 300
    cutoff_freq_high: 8000
    sample_rate: ${sample_rate}
  smooth:
    window_size_range: [2, 10]
  boost_audio:
    amount: 20
  duck_audio:
    amount: 20
  mp3_compression:
    sample_rate: ${sample_rate}
    bitrate: 128k # should be a string e.g. "8k", "32k".. cf ffmpeg to see available bitrates
  aac_compression:
    sample_rate: ${sample_rate}
    bitrate: 128k # should be a string e.g. "8k", "32k".. cf ffmpeg to see available bitrates
    lowpass_freq: null # don't apply low pass freq to ffmpeg aac compression
  encodec:
    ckpt: "//pretrained/facebook/encodec_24khz"
    n_qs: [4, 8, 16]

select_aug_mode:
  "use_eval" # other are 'all' and 'use_eval_acc', used to sample augmentations, `fixed` uses the prob from aug_weights, `all` uses all agmentations every step
  # `use_eval_acc` changes the weights based on the accuracies at evaluation time

aug_weights:
  speed: 0.1
  updownresample: 0.1
  echo: 0.1
  pink_noise: 0.1
  lowpass_filter: 0.1
  highpass_filter: 0.1
  bandpass_filter: 0.1
  smooth: 0.1
  boost_audio: 0.1
  duck_audio: 0.1
  mp3_compression: 0.1 # eval only never use in training even if eval_acc low
  aac_compression: 0.1 # eval only never use in training even if eval_acc low
  encodec: 0.1
  identity: 1 # no augmentation

n_max_aug: null