model:
  name: ecapa_tdnn
  embed_dim: 192
  channels: 512
  se_channels: 128
  res2net_scale: 8
  num_blocks: 3
  n_mels: 128
  aam_m: 0.3
  aam_s: 30.0
  use_band_split: true
  band_split_channels: 128

data:
  sample_rate: 16000
  n_fft: 512
  hop_length: 160
  segment_dur: 5.0
  crop_per_song: 4

training:
  batch_size: 32
  epochs: 50
  lr: 0.001
  weight_decay: 0.0001
  warmup_epochs: 5
  temperature: 0.07
  supcon_weight: 1.0
  aam_weight: 0.3
  mixed_precision: true
  gradient_clip: 1.0
  save_every: 10
  log_every: 10
  sample_type_weights:
    default: 1
    humming_like: 3
    confused: 5
  pair_type_weights:
    default: 1.0
    augmented: 1.4
    humming_like: 2.5
    confused: 4.0

engine:
  chromaprint:
    enabled: true
    n_fft: 1024
    hop_length: 256
  hybrid:
    chroma_weight: 0.25
    ecapa_weight: 0.5
    melody_weight: 0.25
    reject_threshold: 0.4

augmentation:
  pro_wgan_balance: true
  minority_noise_scale: 0.35
  minority_pitch_shift: 8