reach-vb's picture
reach-vb HF staff
Stereo demo update (#60)
5325fcc
# @package __global__
defaults:
- /solver/default
- /model: score/basic
- override /dset: audio/default
- _self_
solver: diffusion
sample_rate: ???
channels: ???
compression_model_checkpoint: ???
n_q: ??? # number of codebooks to keep
dataset:
batch_size: 128
num_workers: 10
segment_duration: 1
train:
num_samples: 500000
valid:
num_samples: 10000
evaluate:
batch_size: 16
num_samples: 10000
generate:
batch_size: 32
num_samples: 50
segment_duration: 10
audio:
sample_rate: ${sample_rate}
loss:
kind: mse
norm_power: 0.
valid:
every: 1
evaluate:
every: 20
num_workers: 5
metrics:
visqol: false
sisnr: false
rvm: true
generate:
every: 25
num_workers: 5
checkpoint:
save_last: true
save_every: 25
keep_last: 10
keep_every_states: null
optim:
epochs: 20000
updates_per_epoch: 2000
lr: 2e-4
max_norm: 0
optimizer: adam
adam:
betas: [0.9, 0.999]
weight_decay: 0.
ema:
use: true # whether to use EMA or not
updates: 1 # update at every step
device: ${device} # device for EMA, can be put on GPU if more frequent updates
decay: 0.99 # EMA decay value, if null, no EMA is used
processor:
name: multi_band_processor
use: false
n_bands: 8
num_samples: 10_000
power_std: 1.
resampling:
use: false
target_sr: 16000
filter:
use: false
n_bands: 4
idx_band: 0
cutoffs: null
schedule:
repartition: "power"
variable_step_batch: true
beta_t0: 1.0e-5
beta_t1: 2.9e-2
beta_exp: 7.5
num_steps: 1000
variance: 'beta'
clip: 5.
rescale: 1.
n_bands: null
noise_scale: 1.0
metrics:
num_stage: 4