Skip to content

Commit

Permalink
fix(train): allow higher segment size (#351)
Browse files Browse the repository at this point in the history
  • Loading branch information
34j committed Apr 16, 2023
1 parent 24c05d1 commit 09d5a52
Show file tree
Hide file tree
Showing 2 changed files with 71 additions and 124 deletions.
194 changes: 70 additions & 124 deletions src/so_vits_svc_fork/modules/commons.py
Original file line number Diff line number Diff line change
@@ -1,27 +1,79 @@
import math
from __future__ import annotations

import torch
from torch.nn import functional as F
import torch.nn.functional as F
from torch import Tensor


def slice_segments(x: Tensor, starts: Tensor, length: int) -> Tensor:
if length is None:
return x
length = min(length, x.size(-1))
x_slice = torch.zeros((x.size()[:-1] + (length,)), dtype=x.dtype, device=x.device)
ends = starts + length
for i, (start, end) in enumerate(zip(starts, ends)):
# LOG.debug(i, start, end, x.size(), x[i, ..., start:end].size(), x_slice.size())
# x_slice[i, ...] = x[i, ..., start:end] need to pad
# x_slice[i, ..., :end - start] = x[i, ..., start:end] this does not work
x_slice[i, ...] = F.pad(x[i, ..., start:end], (0, max(0, length - x.size(-1))))
return x_slice


def rand_slice_segments_with_pitch(
x: Tensor, f0: Tensor, x_lengths: Tensor | int | None, segment_size: int | None
):
if segment_size is None:
return x, f0, torch.arange(x.size(0), device=x.device)
if x_lengths is None:
x_lengths = x.size(-1) * torch.ones(
x.size(0), dtype=torch.long, device=x.device
)
# slice_starts = (torch.rand(z.size(0), device=z.device) * (z_lengths - segment_size)).long()
slice_starts = (
torch.rand(x.size(0), device=x.device)
* torch.max(
x_lengths - segment_size, torch.zeros_like(x_lengths, device=x.device)
)
).long()
z_slice = slice_segments(x, slice_starts, segment_size)
f0_slice = slice_segments(f0, slice_starts, segment_size)
return z_slice, f0_slice, slice_starts


def slice_2d_segments(x: Tensor, starts: Tensor, length: int) -> Tensor:
batch_size, num_features, seq_len = x.shape
ends = starts + length
idxs = (
torch.arange(seq_len, device=x.device)
.unsqueeze(0)
.unsqueeze(1)
.repeat(batch_size, num_features, 1)
)
mask = (idxs >= starts.unsqueeze(-1).unsqueeze(-1)) & (
idxs < ends.unsqueeze(-1).unsqueeze(-1)
)
return x[mask].reshape(batch_size, num_features, length)


def slice_pitch_segments(x, ids_str, segment_size=4):
ret = torch.zeros_like(x[:, :segment_size])
for i in range(x.size(0)):
idx_str = ids_str[i]
idx_end = idx_str + segment_size
ret[i] = x[i, idx_str:idx_end]
return ret
def slice_1d_segments(x: Tensor, starts: Tensor, length: int) -> Tensor:
batch_size, seq_len = x.shape
ends = starts + length
idxs = torch.arange(seq_len, device=x.device).unsqueeze(0).repeat(batch_size, 1)
mask = (idxs >= starts.unsqueeze(-1)) & (idxs < ends.unsqueeze(-1))
return x[mask].reshape(batch_size, length)


def rand_slice_segments_with_pitch(x, pitch, x_lengths=None, segment_size=4):
b, d, t = x.size()
if x_lengths is None:
x_lengths = t
ids_str_max = x_lengths - segment_size + 1
ids_str = (torch.rand([b]).to(device=x.device) * ids_str_max).to(dtype=torch.long)
ret = slice_segments(x, ids_str, segment_size)
ret_pitch = slice_pitch_segments(pitch, ids_str, segment_size)
return ret, ret_pitch, ids_str
def _slice_segments_v3(x: Tensor, starts: Tensor, length: int) -> Tensor:
shape = x.shape[:-1] + (length,)
ends = starts + length
idxs = torch.arange(x.shape[-1], device=x.device).unsqueeze(0).unsqueeze(0)
unsqueeze_dims = len(shape) - len(
x.shape
) # calculate number of dimensions to unsqueeze
starts = starts.reshape(starts.shape + (1,) * unsqueeze_dims)
ends = ends.reshape(ends.shape + (1,) * unsqueeze_dims)
mask = (idxs >= starts) & (idxs < ends)
return x[mask].reshape(shape)


def init_weights(m, mean=0.0, std=0.01):
Expand All @@ -40,89 +92,6 @@ def convert_pad_shape(pad_shape):
return pad_shape


def intersperse(lst, item):
result = [item] * (len(lst) * 2 + 1)
result[1::2] = lst
return result


def kl_divergence(m_p, logs_p, m_q, logs_q):
"""KL(P||Q)"""
kl = (logs_q - logs_p) - 0.5
kl += (
0.5 * (torch.exp(2.0 * logs_p) + ((m_p - m_q) ** 2)) * torch.exp(-2.0 * logs_q)
)
return kl


def rand_gumbel(shape):
"""Sample from the Gumbel distribution, protect from overflows."""
uniform_samples = torch.rand(shape) * 0.99998 + 0.00001
return -torch.log(-torch.log(uniform_samples))


def rand_gumbel_like(x):
g = rand_gumbel(x.size()).to(dtype=x.dtype, device=x.device)
return g


def slice_segments(x, ids_str, segment_size=4):
ret = torch.zeros_like(x[:, :, :segment_size])
for i in range(x.size(0)):
idx_str = ids_str[i]
idx_end = idx_str + segment_size
ret[i] = x[i, :, idx_str:idx_end]
return ret


def rand_slice_segments(x, x_lengths=None, segment_size=4):
b, d, t = x.size()
if x_lengths is None:
x_lengths = t
ids_str_max = x_lengths - segment_size + 1
ids_str = (torch.rand([b]).to(device=x.device) * ids_str_max).to(dtype=torch.long)
ret = slice_segments(x, ids_str, segment_size)
return ret, ids_str


def rand_spec_segments(x, x_lengths=None, segment_size=4):
b, d, t = x.size()
if x_lengths is None:
x_lengths = t
ids_str_max = x_lengths - segment_size
ids_str = (torch.rand([b]).to(device=x.device) * ids_str_max).to(dtype=torch.long)
ret = slice_segments(x, ids_str, segment_size)
return ret, ids_str


def get_timing_signal_1d(length, channels, min_timescale=1.0, max_timescale=1.0e4):
position = torch.arange(length, dtype=torch.float)
num_timescales = channels // 2
log_timescale_increment = math.log(float(max_timescale) / float(min_timescale)) / (
num_timescales - 1
)
inv_timescales = min_timescale * torch.exp(
torch.arange(num_timescales, dtype=torch.float) * -log_timescale_increment
)
scaled_time = position.unsqueeze(0) * inv_timescales.unsqueeze(1)
signal = torch.cat([torch.sin(scaled_time), torch.cos(scaled_time)], 0)
signal = F.pad(signal, [0, 0, 0, channels % 2])
signal = signal.view(1, channels, length)
return signal


def add_timing_signal_1d(x, min_timescale=1.0, max_timescale=1.0e4):
b, channels, length = x.size()
signal = get_timing_signal_1d(length, channels, min_timescale, max_timescale)
return x + signal.to(dtype=x.dtype, device=x.device)


def cat_timing_signal_1d(x, min_timescale=1.0, max_timescale=1.0e4, axis=1):
b, channels, length = x.size()
signal = get_timing_signal_1d(length, channels, min_timescale, max_timescale)
return torch.cat([x, signal.to(dtype=x.dtype, device=x.device)], axis)


def subsequent_mask(length):
mask = torch.tril(torch.ones(length, length)).unsqueeze(0).unsqueeze(0)
return mask
Expand All @@ -138,36 +107,13 @@ def fused_add_tanh_sigmoid_multiply(input_a, input_b, n_channels):
return acts


def shift_1d(x):
x = F.pad(x, convert_pad_shape([[0, 0], [0, 0], [1, 0]]))[:, :, :-1]
return x


def sequence_mask(length, max_length=None):
if max_length is None:
max_length = length.max()
x = torch.arange(max_length, dtype=length.dtype, device=length.device)
return x.unsqueeze(0) < length.unsqueeze(1)


def generate_path(duration, mask):
"""
duration: [b, 1, t_x]
mask: [b, 1, t_y, t_x]
"""
duration.device

b, _, t_y, t_x = mask.shape
cum_duration = torch.cumsum(duration, -1)

cum_duration_flat = cum_duration.view(b * t_x)
path = sequence_mask(cum_duration_flat, t_y).to(mask.dtype)
path = path.view(b, t_x, t_y)
path = path - F.pad(path, convert_pad_shape([[0, 0], [1, 0], [0, 0]]))[:, :-1]
path = path.unsqueeze(1).transpose(2, 3) * mask
return path


def clip_grad_value_(parameters, clip_value, norm_type=2):
if isinstance(parameters, torch.Tensor):
parameters = [parameters]
Expand Down
1 change: 1 addition & 0 deletions src/so_vits_svc_fork/train.py
Original file line number Diff line number Diff line change
Expand Up @@ -375,6 +375,7 @@ def training_step(self, batch: dict[str, torch.Tensor], batch_idx: int) -> None:
ids_slice * self.hparams.data.hop_length,
self.hparams.train.segment_size,
)
y = y[..., : y_hat.shape[-1]]

# generator loss
y_d_hat_r, y_d_hat_g, fmap_r, fmap_g = self.net_d(y, y_hat)
Expand Down

0 comments on commit 09d5a52

Please sign in to comment.