xjsc0's picture
1
64ec292
audio:
chunk_size: 352800
dim_f: 1024
dim_t: 801 # don't work (use in model)
hop_length: 441 # don't work (use in model)
n_fft: 2048
num_channels: 2
sample_rate: 44100
min_mean_abs: 0.000
model:
dim: 256
depth: 6
stereo: true
time_module_depth: 1
freq_module_depth: 1
mask_estimator_depth: 2
module_type: mamba2
dim_head: 48
heads: 8
ff_dropout: 0.1
attn_dropout: 0.1
flash_attn: true
mamba_gmlp: false
num_stems: 1
stft_n_fft: 2048
stft_hop_length: 441
stft_win_length: 2048
stft_normalized: false
multi_stft_resolution_loss_weight: 1.0
multi_stft_resolutions_window_sizes: !!python/tuple
- 4096
- 2048
- 1024
- 512
- 256
multi_stft_hop_size: 147
multi_stft_normalized: false
training:
batch_size: 6
gradient_accumulation_steps: 1
grad_clip: 0
instruments: ['vocals', 'other']
patience: 2
reduce_factor: 0.95
target_instrument: vocals
num_epochs: 1000
num_steps: 1000
q: 0.95
coarse_loss_clip: true
ema_momentum: 0.999
# optimizer: prodigy
# lr: 1.0
optimizer: adam
lr: 1.0e-5
other_fix: true # it's needed for checking on multisong dataset if other is actually instrumental
use_amp: true # enable or disable usage of mixed precision (float16) - usually it must be true
augmentations:
enable: true # enable or disable all augmentations (to fast disable if needed)
loudness: true # randomly change loudness of each stem on the range (loudness_min; loudness_max)
loudness_min: 0.5
loudness_max: 1.5
mixup: true # mix several stems of same type with some probability (only works for dataset types: 1, 2, 3)
mixup_probs: !!python/tuple # 2 additional stems of the same type (1st with prob 0.2, 2nd with prob 0.02)
- 0.2
- 0.02
- 0.002
mixup_loudness_min: 0.5
mixup_loudness_max: 1.5
all:
channel_shuffle: 0.5 # Set 0 or lower to disable
random_inverse: 0.1 # inverse track (better lower probability)
random_polarity: 0.5 # polarity change (multiply waveform to -1)
vocals:
pitch_shift: 0.01
pitch_shift_min_semitones: -5
pitch_shift_max_semitones: 5
seven_band_parametric_eq: 0.01
seven_band_parametric_eq_min_gain_db: -9
seven_band_parametric_eq_max_gain_db: 9
tanh_distortion: 0.01
tanh_distortion_min: 0.1
tanh_distortion_max: 0.7
other:
pitch_shift: 0.01
pitch_shift_min_semitones: -4
pitch_shift_max_semitones: 4
gaussian_noise: 0.01
gaussian_noise_min_amplitude: 0.001
gaussian_noise_max_amplitude: 0.015
time_stretch: 0.01
time_stretch_min_rate: 0.8
time_stretch_max_rate: 1.25
inference:
chunk_size: 352800
batch_size: 4
dim_t: 1101
num_overlap: 2
normalize: false