Spaces:
Sleeping
Sleeping
File size: 1,055 Bytes
150fe29 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 |
from fish_diffusion.datasets.hifisinger import HiFiSVCDataset
from fish_diffusion.datasets.utils import get_datasets_from_subfolder
_base_ = [
"./_base_/archs/hifi_svc.py",
"./_base_/trainers/base.py",
"./_base_/schedulers/exponential.py",
"./_base_/datasets/hifi_svc.py",
]
speaker_mapping = {
"Placeholder": 0,
}
model = dict(
type="HiFiSVC",
speaker_encoder=dict(
input_size=len(speaker_mapping),
),
)
preprocessing = dict(
text_features_extractor=dict(
type="ContentVec",
),
pitch_extractor=dict(
type="CrepePitchExtractor",
keep_zeros=False,
f0_min=40.0,
f0_max=2000.0,
),
energy_extractor=dict(
type="RMSEnergyExtractor",
),
augmentations=[
dict(
type="FixedPitchShifting",
key_shifts=[-5.0, 5.0],
probability=0.75,
),
],
)
trainer = dict(
# Disable gradient clipping, which is not supported by custom optimization
gradient_clip_val=None,
max_steps=1000000,
)
|