Nik Afiq 5238298b55
Some checks failed
CI / test (push) Successful in 6s
CI / build-ai-gateway (push) Failing after 28s
CI / build-ha-gateway (push) Failing after 24s
CI / build-discord-bot (push) Failing after 29s
Add TTS model components and inference server
- Implemented core model components in `modules.py` including various convolutional layers and normalization techniques.
- Added transformation functions in `transforms.py` for piecewise rational quadratic transformations.
- Created utility functions in `utils.py` for checkpoint management, logging, and hyperparameter handling.
- Introduced monotonic alignment functionality with Cython optimization in `monotonic_align`.
- Developed a minimal inference server in `server.py` to handle synthesis requests.
- Updated requirements to include necessary dependencies for Cython and scipy.
2026-07-24 22:38:36 +09:00

210 lines
9.6 KiB
Python

import torch
from torch.nn import functional as F
import numpy as np
DEFAULT_MIN_BIN_WIDTH = 1e-3
DEFAULT_MIN_BIN_HEIGHT = 1e-3
DEFAULT_MIN_DERIVATIVE = 1e-3
def piecewise_rational_quadratic_transform(inputs,
unnormalized_widths,
unnormalized_heights,
unnormalized_derivatives,
inverse=False,
tails=None,
tail_bound=1.,
min_bin_width=DEFAULT_MIN_BIN_WIDTH,
min_bin_height=DEFAULT_MIN_BIN_HEIGHT,
min_derivative=DEFAULT_MIN_DERIVATIVE):
if tails is None:
spline_fn = rational_quadratic_spline
spline_kwargs = {}
else:
spline_fn = unconstrained_rational_quadratic_spline
spline_kwargs = {
'tails': tails,
'tail_bound': tail_bound
}
outputs, logabsdet = spline_fn(
inputs=inputs,
unnormalized_widths=unnormalized_widths,
unnormalized_heights=unnormalized_heights,
unnormalized_derivatives=unnormalized_derivatives,
inverse=inverse,
min_bin_width=min_bin_width,
min_bin_height=min_bin_height,
min_derivative=min_derivative,
**spline_kwargs
)
return outputs, logabsdet
def searchsorted(bin_locations, inputs, eps=1e-6):
bin_locations[..., -1] += eps
return torch.sum(
inputs[..., None] >= bin_locations,
dim=-1
) - 1
def unconstrained_rational_quadratic_spline(inputs,
unnormalized_widths,
unnormalized_heights,
unnormalized_derivatives,
inverse=False,
tails='linear',
tail_bound=1.,
min_bin_width=DEFAULT_MIN_BIN_WIDTH,
min_bin_height=DEFAULT_MIN_BIN_HEIGHT,
min_derivative=DEFAULT_MIN_DERIVATIVE):
inside_interval_mask = (inputs >= -tail_bound) & (inputs <= tail_bound)
if tails == 'linear':
unnormalized_derivatives = F.pad(unnormalized_derivatives, pad=(1, 1))
constant = np.log(np.exp(1 - min_derivative) - 1)
unnormalized_derivatives[..., 0] = constant
unnormalized_derivatives[..., -1] = constant
else:
raise RuntimeError('{} tails are not implemented.'.format(tails))
# ONNX-export-friendly rewrite (Phase 0, TTS_GATEWAY_PLAN.md): the original
# boolean-mask indexed assignment (`outputs[mask] = ...`) subselects a
# data-dependent number of elements, which torch.export's symbolic tracer
# can't handle (unbacked SymInts leaking into later Conv1d shape guards).
# Rewritten as a fixed-shape computation + torch.where select instead.
# Clamping is a no-op wherever inside_interval_mask is True (those inputs
# are already within bounds), so the spline output there is identical to
# the original masked version; where the mask is False, the spline's
# output for that clamped position is simply discarded by torch.where in
# favor of the identity passthrough - clamping there only exists to keep
# rational_quadratic_spline's internal `gather` calls in-bounds.
clamped_inputs = torch.clamp(inputs, -tail_bound, tail_bound)
spline_outputs, spline_logabsdet = rational_quadratic_spline(
inputs=clamped_inputs,
unnormalized_widths=unnormalized_widths,
unnormalized_heights=unnormalized_heights,
unnormalized_derivatives=unnormalized_derivatives,
inverse=inverse,
left=-tail_bound, right=tail_bound, bottom=-tail_bound, top=tail_bound,
min_bin_width=min_bin_width,
min_bin_height=min_bin_height,
min_derivative=min_derivative
)
outputs = torch.where(inside_interval_mask, spline_outputs, inputs)
logabsdet = torch.where(inside_interval_mask, spline_logabsdet, torch.zeros_like(inputs))
return outputs, logabsdet
def rational_quadratic_spline(inputs,
unnormalized_widths,
unnormalized_heights,
unnormalized_derivatives,
inverse=False,
left=0., right=1., bottom=0., top=1.,
min_bin_width=DEFAULT_MIN_BIN_WIDTH,
min_bin_height=DEFAULT_MIN_BIN_HEIGHT,
min_derivative=DEFAULT_MIN_DERIVATIVE):
# Removed for ONNX export (Phase 0 of TTS_GATEWAY_PLAN.md): this is a pure
# bounds assertion on tensor values, not part of the computation - the
# caller (unconstrained_rational_quadratic_spline) already only invokes
# this on inputs[inside_interval_mask], which are guaranteed in [left,
# right] by construction. Left in as a data-dependent Python `if`, it
# breaks torch.export's symbolic tracer (GuardOnDataDependentSymNode).
# Behavior is unchanged either way; only affects traceability.
num_bins = unnormalized_widths.shape[-1]
if min_bin_width * num_bins > 1.0:
raise ValueError('Minimal bin width too large for the number of bins')
if min_bin_height * num_bins > 1.0:
raise ValueError('Minimal bin height too large for the number of bins')
widths = F.softmax(unnormalized_widths, dim=-1)
widths = min_bin_width + (1 - min_bin_width * num_bins) * widths
cumwidths = torch.cumsum(widths, dim=-1)
cumwidths = F.pad(cumwidths, pad=(1, 0), mode='constant', value=0.0)
cumwidths = (right - left) * cumwidths + left
cumwidths[..., 0] = left
cumwidths[..., -1] = right
widths = cumwidths[..., 1:] - cumwidths[..., :-1]
derivatives = min_derivative + F.softplus(unnormalized_derivatives)
heights = F.softmax(unnormalized_heights, dim=-1)
heights = min_bin_height + (1 - min_bin_height * num_bins) * heights
cumheights = torch.cumsum(heights, dim=-1)
cumheights = F.pad(cumheights, pad=(1, 0), mode='constant', value=0.0)
cumheights = (top - bottom) * cumheights + bottom
cumheights[..., 0] = bottom
cumheights[..., -1] = top
heights = cumheights[..., 1:] - cumheights[..., :-1]
if inverse:
bin_idx = searchsorted(cumheights, inputs)[..., None]
else:
bin_idx = searchsorted(cumwidths, inputs)[..., None]
input_cumwidths = cumwidths.gather(-1, bin_idx)[..., 0]
input_bin_widths = widths.gather(-1, bin_idx)[..., 0]
input_cumheights = cumheights.gather(-1, bin_idx)[..., 0]
delta = heights / widths
input_delta = delta.gather(-1, bin_idx)[..., 0]
input_derivatives = derivatives.gather(-1, bin_idx)[..., 0]
input_derivatives_plus_one = derivatives[..., 1:].gather(-1, bin_idx)[..., 0]
input_heights = heights.gather(-1, bin_idx)[..., 0]
if inverse:
a = (((inputs - input_cumheights) * (input_derivatives
+ input_derivatives_plus_one
- 2 * input_delta)
+ input_heights * (input_delta - input_derivatives)))
b = (input_heights * input_derivatives
- (inputs - input_cumheights) * (input_derivatives
+ input_derivatives_plus_one
- 2 * input_delta))
c = - input_delta * (inputs - input_cumheights)
discriminant = b.pow(2) - 4 * a * c
# Removed for ONNX export (Phase 0 of TTS_GATEWAY_PLAN.md) - same
# reasoning as the bounds check above: a numerical-invariant sanity
# check on a tensor value, not part of the computation itself, and it
# breaks torch.export's symbolic tracer the same way.
root = (2 * c) / (-b - torch.sqrt(discriminant))
outputs = root * input_bin_widths + input_cumwidths
theta_one_minus_theta = root * (1 - root)
denominator = input_delta + ((input_derivatives + input_derivatives_plus_one - 2 * input_delta)
* theta_one_minus_theta)
derivative_numerator = input_delta.pow(2) * (input_derivatives_plus_one * root.pow(2)
+ 2 * input_delta * theta_one_minus_theta
+ input_derivatives * (1 - root).pow(2))
logabsdet = torch.log(derivative_numerator) - 2 * torch.log(denominator)
return outputs, -logabsdet
else:
theta = (inputs - input_cumwidths) / input_bin_widths
theta_one_minus_theta = theta * (1 - theta)
numerator = input_heights * (input_delta * theta.pow(2)
+ input_derivatives * theta_one_minus_theta)
denominator = input_delta + ((input_derivatives + input_derivatives_plus_one - 2 * input_delta)
* theta_one_minus_theta)
outputs = input_cumheights + numerator / denominator
derivative_numerator = input_delta.pow(2) * (input_derivatives_plus_one * theta.pow(2)
+ 2 * input_delta * theta_one_minus_theta
+ input_derivatives * (1 - theta).pow(2))
logabsdet = torch.log(derivative_numerator) - 2 * torch.log(denominator)
return outputs, logabsdet