You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
328 lines
13 KiB
328 lines
13 KiB
# Copyright (c) 2023 PaddlePaddle Authors. All Rights Reserved.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
"""Diffusion denoising related modules for paddle"""
|
|
import math
|
|
from typing import Callable
|
|
from typing import Optional
|
|
from typing import Tuple
|
|
|
|
import numpy as np
|
|
import paddle
|
|
import ppdiffusers
|
|
from paddle import nn
|
|
from ppdiffusers.models.embeddings import Timesteps
|
|
from ppdiffusers.schedulers import DDPMScheduler
|
|
|
|
from paddlespeech.t2s.modules.nets_utils import initialize
|
|
from paddlespeech.t2s.modules.residual_block import WaveNetResidualBlock
|
|
|
|
|
|
class GaussianDiffusion(nn.Layer):
|
|
"""Common Gaussian Diffusion Denoising Model Module
|
|
|
|
Args:
|
|
denoiser (Layer, optional):
|
|
The model used for denoising noises.
|
|
num_train_timesteps (int, optional):
|
|
The number of timesteps between the noise and the real during training, by default 1000.
|
|
beta_start (float, optional):
|
|
beta start parameter for the scheduler, by default 0.0001.
|
|
beta_end (float, optional):
|
|
beta end parameter for the scheduler, by default 0.0001.
|
|
beta_schedule (str, optional):
|
|
beta schedule parameter for the scheduler, by default 'squaredcos_cap_v2' (cosine schedule).
|
|
num_max_timesteps (int, optional):
|
|
The max timestep transition from real to noise, by default None.
|
|
stretch (bool, optional):
|
|
Whether to stretch before diffusion, by defalut True.
|
|
min_values: (paddle.Tensor):
|
|
The minimum value of the feature to stretch.
|
|
max_values: (paddle.Tensor):
|
|
The maximum value of the feature to stretch.
|
|
|
|
|
|
Examples:
|
|
>>> import paddle
|
|
>>> import paddle.nn.functional as F
|
|
>>> from tqdm import tqdm
|
|
>>>
|
|
>>> denoiser = WaveNetDenoiser()
|
|
>>> diffusion = GaussianDiffusion(denoiser, num_train_timesteps=1000, num_max_timesteps=100)
|
|
>>> x = paddle.ones([4, 80, 192]) # [B, mel_ch, T] # real mel input
|
|
>>> c = paddle.randn([4, 256, 192]) # [B, fs2_encoder_out_ch, T] # fastspeech2 encoder output
|
|
>>> loss = F.mse_loss(*diffusion(x, c))
|
|
>>> loss.backward()
|
|
>>> print('MSE Loss:', loss.item())
|
|
MSE Loss: 1.6669728755950928
|
|
>>> def create_progress_callback():
|
|
>>> pbar = None
|
|
>>> def callback(index, timestep, num_timesteps, sample):
|
|
>>> nonlocal pbar
|
|
>>> if pbar is None:
|
|
>>> pbar = tqdm(total=num_timesteps)
|
|
>>> pbar.update(index)
|
|
>>> pbar.update()
|
|
>>>
|
|
>>> return callback
|
|
>>>
|
|
>>> # ds=1000, K_step=60, scheduler=ddpm, from aux fs2 mel output
|
|
>>> ds = 1000
|
|
>>> infer_steps = 1000
|
|
>>> K_step = 60
|
|
>>> scheduler_type = 'ddpm'
|
|
>>> x_in = x
|
|
>>> diffusion = GaussianDiffusion(denoiser, num_train_timesteps=ds, num_max_timesteps=K_step)
|
|
>>> with paddle.no_grad():
|
|
>>> sample = diffusion.inference(
|
|
>>> paddle.randn(x.shape), c, ref_x=x_in,
|
|
>>> num_inference_steps=infer_steps,
|
|
>>> scheduler_type=scheduler_type,
|
|
>>> callback=create_progress_callback())
|
|
100%|█████| 60/60 [00:03<00:00, 18.36it/s]
|
|
>>>
|
|
>>> # ds=100, K_step=100, scheduler=ddpm, from gaussian noise
|
|
>>> ds = 100
|
|
>>> infer_steps = 100
|
|
>>> K_step = 100
|
|
>>> scheduler_type = 'ddpm'
|
|
>>> x_in = None
|
|
>>> diffusion = GaussianDiffusion(denoiser, num_train_timesteps=ds, num_max_timesteps=K_step)
|
|
>>> with paddle.no_grad():
|
|
>>> sample = diffusion.inference(
|
|
>>> paddle.randn(x.shape), c, ref_x=x_in,
|
|
>>> num_inference_steps=infer_steps,
|
|
>>> scheduler_type=scheduler_type,
|
|
>>> callback=create_progress_callback())
|
|
100%|█████| 100/100 [00:05<00:00, 18.29it/s]
|
|
>>>
|
|
>>> # ds=1000, K_step=1000, scheduler=pndm, infer_step=25, from gaussian noise
|
|
>>> ds = 1000
|
|
>>> infer_steps = 25
|
|
>>> K_step = 1000
|
|
>>> scheduler_type = 'pndm'
|
|
>>> x_in = None
|
|
>>> diffusion = GaussianDiffusion(denoiser, num_train_timesteps=ds, num_max_timesteps=K_step)
|
|
>>> with paddle.no_grad():
|
|
>>> sample = diffusion.inference(
|
|
>>> paddle.randn(x.shape), c, ref_x=x_in,
|
|
>>> num_inference_steps=infer_steps,
|
|
>>> scheduler_type=scheduler_type,
|
|
>>> callback=create_progress_callback())
|
|
100%|█████| 34/34 [00:01<00:00, 19.75it/s]
|
|
>>>
|
|
>>> # ds=1000, K_step=100, scheduler=pndm, infer_step=50, from aux fs2 mel output
|
|
>>> ds = 1000
|
|
>>> infer_steps = 50
|
|
>>> K_step = 100
|
|
>>> scheduler_type = 'pndm'
|
|
>>> x_in = x
|
|
>>> diffusion = GaussianDiffusion(denoiser, num_train_timesteps=ds, num_max_timesteps=K_step)
|
|
>>> with paddle.no_grad():
|
|
>>> sample = diffusion.inference(
|
|
>>> paddle.randn(x.shape), c, ref_x=x_in,
|
|
>>> num_inference_steps=infer_steps,
|
|
>>> scheduler_type=scheduler_type,
|
|
>>> callback=create_progress_callback())
|
|
100%|█████| 14/14 [00:00<00:00, 23.80it/s]
|
|
|
|
"""
|
|
|
|
def __init__(
|
|
self,
|
|
denoiser: nn.Layer,
|
|
num_train_timesteps: Optional[int]=1000,
|
|
beta_start: Optional[float]=0.0001,
|
|
beta_end: Optional[float]=0.02,
|
|
beta_schedule: Optional[str]="squaredcos_cap_v2",
|
|
num_max_timesteps: Optional[int]=None,
|
|
stretch: bool=True,
|
|
min_values: paddle.Tensor=None,
|
|
max_values: paddle.Tensor=None, ):
|
|
super().__init__()
|
|
|
|
self.num_train_timesteps = num_train_timesteps
|
|
self.beta_start = beta_start
|
|
self.beta_end = beta_end
|
|
self.beta_schedule = beta_schedule
|
|
|
|
self.denoiser = denoiser
|
|
self.noise_scheduler = DDPMScheduler(
|
|
num_train_timesteps=num_train_timesteps,
|
|
beta_start=beta_start,
|
|
beta_end=beta_end,
|
|
beta_schedule=beta_schedule)
|
|
self.num_max_timesteps = num_max_timesteps
|
|
self.stretch = stretch
|
|
self.min_values = min_values
|
|
self.max_values = max_values
|
|
|
|
def norm_spec(self, x):
|
|
"""
|
|
Linearly map x to [-1, 1]
|
|
Args:
|
|
x: [B, T, N]
|
|
"""
|
|
return (x - self.min_values) / (self.max_values - self.min_values
|
|
) * 2 - 1
|
|
|
|
def denorm_spec(self, x):
|
|
return (x + 1) / 2 * (self.max_values - self.min_values
|
|
) + self.min_values
|
|
|
|
def forward(self, x: paddle.Tensor, cond: Optional[paddle.Tensor]=None
|
|
) -> Tuple[paddle.Tensor, paddle.Tensor]:
|
|
"""Generate random timesteps noised x.
|
|
|
|
Args:
|
|
x (Tensor):
|
|
The input for adding noises.
|
|
cond (Tensor, optional):
|
|
Conditional input for compute noises.
|
|
|
|
Returns:
|
|
y (Tensor):
|
|
The output with noises added in.
|
|
target (Tensor):
|
|
The noises which is added to the input.
|
|
|
|
"""
|
|
if self.stretch:
|
|
x = x.transpose((0, 2, 1))
|
|
x = self.norm_spec(x)
|
|
x = x.transpose((0, 2, 1))
|
|
|
|
noise_scheduler = self.noise_scheduler
|
|
|
|
# Sample noise that we'll add to the mel-spectrograms
|
|
target = noise = paddle.randn(x.shape)
|
|
|
|
# Sample a random timestep for each mel-spectrogram
|
|
num_timesteps = self.num_train_timesteps
|
|
if self.num_max_timesteps is not None:
|
|
num_timesteps = self.num_max_timesteps
|
|
timesteps = paddle.randint(0, num_timesteps, (x.shape[0], ))
|
|
|
|
# Add noise to the clean mel-spectrograms according to the noise magnitude at each timestep
|
|
# (this is the forward diffusion process)
|
|
noisy_images = noise_scheduler.add_noise(x, noise, timesteps)
|
|
|
|
y = self.denoiser(noisy_images, timesteps, cond)
|
|
|
|
# then compute loss use output y and noisy target for prediction_type == "epsilon"
|
|
return y, target
|
|
|
|
def inference(self,
|
|
noise: paddle.Tensor,
|
|
cond: Optional[paddle.Tensor]=None,
|
|
ref_x: Optional[paddle.Tensor]=None,
|
|
num_inference_steps: Optional[int]=1000,
|
|
strength: Optional[float]=None,
|
|
scheduler_type: Optional[str]="ddpm",
|
|
clip_noise: Optional[bool]=False,
|
|
clip_noise_range: Optional[Tuple[float, float]]=(-1, 1),
|
|
callback: Optional[Callable[[int, int, int, paddle.Tensor],
|
|
None]]=None,
|
|
callback_steps: Optional[int]=1):
|
|
"""Denoising input from noises. Refer to ppdiffusers img2img pipeline.
|
|
|
|
Args:
|
|
noise (Tensor):
|
|
The input tensor as a starting point for denoising.
|
|
cond (Tensor, optional):
|
|
Conditional input for compute noises. (N, C_aux, T)
|
|
ref_x (Tensor, optional):
|
|
The real output for the denoising process to refer.
|
|
num_inference_steps (int, optional):
|
|
The number of timesteps between the noise and the real during inference, by default 1000.
|
|
strength (float, optional):
|
|
Mixing strength of ref_x with noise. The larger the value, the stronger the noise.
|
|
Range [0,1], by default None.
|
|
scheduler_type (str, optional):
|
|
Noise scheduler for generate noises.
|
|
Choose a great scheduler can skip many denoising step, by default 'ddpm'.
|
|
only support 'ddpm' now !
|
|
clip_noise (bool, optional):
|
|
Whether to clip each denoised output, by default True.
|
|
clip_noise_range (tuple, optional):
|
|
denoised output min and max value range after clip, by default (-1, 1).
|
|
callback (Callable[[int,int,int,Tensor], None], optional):
|
|
Callback function during denoising steps.
|
|
|
|
Args:
|
|
index (int):
|
|
Current denoising index.
|
|
timestep (int):
|
|
Current denoising timestep.
|
|
num_timesteps (int):
|
|
Number of the denoising timesteps.
|
|
denoised_output (Tensor):
|
|
Current intermediate result produced during denoising.
|
|
|
|
callback_steps (int, optional):
|
|
The step to call the callback function.
|
|
|
|
Returns:
|
|
denoised_output (Tensor):
|
|
The denoised output tensor.
|
|
|
|
"""
|
|
scheduler_cls = None
|
|
for clsname in dir(ppdiffusers.schedulers):
|
|
if clsname.lower() == scheduler_type + "scheduler":
|
|
scheduler_cls = getattr(ppdiffusers.schedulers, clsname)
|
|
break
|
|
|
|
if scheduler_cls is None:
|
|
raise ValueError(f"No such scheduler type named {scheduler_type}")
|
|
|
|
scheduler = scheduler_cls(
|
|
num_train_timesteps=self.num_train_timesteps,
|
|
beta_start=self.beta_start,
|
|
beta_end=self.beta_end,
|
|
beta_schedule=self.beta_schedule)
|
|
|
|
# set timesteps
|
|
scheduler.set_timesteps(num_inference_steps)
|
|
|
|
noisy_input = noise
|
|
if self.stretch and ref_x is not None:
|
|
ref_x = ref_x.transpose((0, 2, 1))
|
|
ref_x = self.norm_spec(ref_x)
|
|
ref_x = ref_x.transpose((0, 2, 1))
|
|
|
|
# for ddpm
|
|
timesteps = paddle.to_tensor(
|
|
np.flipud(np.arange(num_inference_steps)))
|
|
noisy_input = scheduler.add_noise(ref_x, noise, timesteps[0])
|
|
|
|
denoised_output = noisy_input
|
|
if clip_noise:
|
|
n_min, n_max = clip_noise_range
|
|
denoised_output = paddle.clip(denoised_output, n_min, n_max)
|
|
for i, t in enumerate(timesteps):
|
|
denoised_output = scheduler.scale_model_input(denoised_output, t)
|
|
noise_pred = self.denoiser(denoised_output, t, cond)
|
|
# compute the previous noisy sample x_t -> x_t-1
|
|
denoised_output = scheduler.step(noise_pred, t,
|
|
denoised_output).prev_sample
|
|
if clip_noise:
|
|
denoised_output = paddle.clip(denoised_output, n_min, n_max)
|
|
|
|
if self.stretch:
|
|
denoised_output = denoised_output.transpose((0, 2, 1))
|
|
denoised_output = self.denorm_spec(denoised_output)
|
|
denoised_output = denoised_output.transpose((0, 2, 1))
|
|
|
|
return denoised_output
|