313 lines
13 KiB
Python
313 lines
13 KiB
Python
import functools
|
|
import random
|
|
|
|
import torch
|
|
import torch.nn as nn
|
|
import torch.nn.functional as F
|
|
from torch import autocast
|
|
|
|
from models.diffusion.nn import timestep_embedding, normalization, zero_module, conv_nd, linear
|
|
from models.diffusion.unet_diffusion import TimestepEmbedSequential, \
|
|
Downsample, Upsample
|
|
from models.audio.tts.mini_encoder import AudioMiniEncoder
|
|
from scripts.audio.gen.use_diffuse_tts import ceil_multiple
|
|
from trainer.networks import register_model
|
|
from x_transformers import Encoder, ContinuousTransformerWrapper
|
|
|
|
|
|
class CheckpointedLayer(nn.Module):
|
|
"""
|
|
Wraps a module. When forward() is called, passes kwargs that require_grad through torch.checkpoint() and bypasses
|
|
checkpoint for all other args.
|
|
"""
|
|
def __init__(self, wrap):
|
|
super().__init__()
|
|
self.wrap = wrap
|
|
|
|
def forward(self, x, *args, **kwargs):
|
|
for k, v in kwargs.items():
|
|
assert not (isinstance(v, torch.Tensor) and v.requires_grad) # This would screw up checkpointing.
|
|
partial = functools.partial(self.wrap, **kwargs)
|
|
return torch.utils.checkpoint.checkpoint(partial, x, *args)
|
|
|
|
|
|
class CheckpointedXTransformerEncoder(nn.Module):
|
|
"""
|
|
Wraps a ContinuousTransformerWrapper and applies CheckpointedLayer to each layer and permutes from channels-mid
|
|
to channels-last that XTransformer expects.
|
|
"""
|
|
def __init__(self, **xtransformer_kwargs):
|
|
super().__init__()
|
|
self.transformer = ContinuousTransformerWrapper(**xtransformer_kwargs)
|
|
|
|
for i in range(len(self.transformer.attn_layers.layers)):
|
|
n, b, r = self.transformer.attn_layers.layers[i]
|
|
self.transformer.attn_layers.layers[i] = nn.ModuleList([n, CheckpointedLayer(b), r])
|
|
|
|
def forward(self, x, **kwargs):
|
|
x = x.permute(0,2,1)
|
|
h = self.transformer(x, **kwargs)
|
|
return h.permute(0,2,1)
|
|
|
|
|
|
class DiffusionTts(nn.Module):
|
|
def __init__(
|
|
self,
|
|
model_channels,
|
|
in_channels=1,
|
|
num_tokens=32,
|
|
out_channels=2, # mean and variance
|
|
dropout=0,
|
|
# res 1, 2, 4, 8,16,32,64,128,256,512, 1K, 2K
|
|
channel_mult= (1,1.5,2, 3, 4, 6, 8, 12, 16, 24, 32, 48),
|
|
# spec_cond: 1, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0)
|
|
# attn: 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1
|
|
token_conditioning_resolutions=(1,16,),
|
|
dims=1,
|
|
use_fp16=False,
|
|
time_embed_dim_multiplier=4,
|
|
cond_transformer_depth=8,
|
|
mid_transformer_depth=8,
|
|
nil_guidance_fwd_proportion=.3,
|
|
# Parameters for super-sampling.
|
|
super_sampling=False,
|
|
super_sampling_max_noising_factor=.1,
|
|
# Parameters for unaligned inputs.
|
|
enabled_unaligned_inputs=False,
|
|
num_unaligned_tokens=164,
|
|
unaligned_encoder_depth=8,
|
|
):
|
|
super().__init__()
|
|
|
|
if super_sampling:
|
|
in_channels *= 2 # In super-sampling mode, the LR input is concatenated directly onto the input.
|
|
self.in_channels = in_channels
|
|
self.model_channels = model_channels
|
|
self.out_channels = out_channels
|
|
self.dropout = dropout
|
|
self.channel_mult = channel_mult
|
|
self.dtype = torch.float16 if use_fp16 else torch.float32
|
|
self.dims = dims
|
|
self.nil_guidance_fwd_proportion = nil_guidance_fwd_proportion
|
|
self.mask_token_id = num_tokens
|
|
self.super_sampling_enabled = super_sampling
|
|
self.super_sampling_max_noising_factor = super_sampling_max_noising_factor
|
|
|
|
time_embed_dim = model_channels * time_embed_dim_multiplier
|
|
self.time_embed = nn.Sequential(
|
|
linear(model_channels, time_embed_dim),
|
|
nn.SiLU(),
|
|
linear(time_embed_dim, time_embed_dim),
|
|
)
|
|
|
|
embedding_dim = model_channels * 8
|
|
self.code_embedding = nn.Embedding(num_tokens+1, embedding_dim)
|
|
self.contextual_embedder = AudioMiniEncoder(1, embedding_dim, base_channels=32, depth=6, resnet_blocks=1,
|
|
attn_blocks=2, num_attn_heads=2, dropout=dropout, downsample_factor=4, kernel_size=5)
|
|
self.conditioning_conv = nn.Conv1d(embedding_dim*3, embedding_dim, 1)
|
|
|
|
self.enable_unaligned_inputs = enabled_unaligned_inputs
|
|
if enabled_unaligned_inputs:
|
|
self.unaligned_embedder = nn.Embedding(num_unaligned_tokens, embedding_dim)
|
|
self.unaligned_encoder = CheckpointedXTransformerEncoder(
|
|
max_seq_len=-1,
|
|
use_pos_emb=False,
|
|
attn_layers=Encoder(
|
|
dim=embedding_dim,
|
|
depth=unaligned_encoder_depth,
|
|
heads=embedding_dim//128,
|
|
ff_dropout=dropout,
|
|
attn_dropout=dropout,
|
|
use_rmsnorm=True,
|
|
ff_glu=True,
|
|
rotary_emb_dim=True,
|
|
)
|
|
)
|
|
|
|
self.conditioning_encoder = CheckpointedXTransformerEncoder(
|
|
max_seq_len=-1, # Should be unused
|
|
use_pos_emb=False,
|
|
attn_layers=Encoder(
|
|
dim=embedding_dim,
|
|
depth=cond_transformer_depth,
|
|
heads=embedding_dim//128,
|
|
ff_dropout=dropout,
|
|
attn_dropout=dropout,
|
|
use_rmsnorm=True,
|
|
ff_glu=True,
|
|
rotary_pos_emb=True,
|
|
cross_attend=self.enable_unaligned_inputs,
|
|
)
|
|
)
|
|
|
|
self.input_blocks = nn.ModuleList(
|
|
[
|
|
TimestepEmbedSequential(
|
|
conv_nd(dims, in_channels, model_channels, 3, padding=1)
|
|
)
|
|
]
|
|
)
|
|
token_conditioning_blocks = []
|
|
input_block_chans = [model_channels]
|
|
ch = model_channels
|
|
ds = 1
|
|
|
|
for level, mult in enumerate(channel_mult):
|
|
if ds in token_conditioning_resolutions:
|
|
token_conditioning_block = nn.Conv1d(embedding_dim, ch, 1)
|
|
token_conditioning_block.weight.data *= .02
|
|
self.input_blocks.append(token_conditioning_block)
|
|
token_conditioning_blocks.append(token_conditioning_block)
|
|
|
|
out_ch = int(mult * model_channels)
|
|
if level != len(channel_mult) - 1:
|
|
self.input_blocks.append(
|
|
TimestepEmbedSequential(
|
|
Downsample(
|
|
ch, use_conv=True, dims=dims, out_channels=out_ch, factor=2, ksize=3, pad=1
|
|
)
|
|
)
|
|
)
|
|
ch = out_ch
|
|
input_block_chans.append(ch)
|
|
ds *= 2
|
|
|
|
self.middle_block = CheckpointedXTransformerEncoder(
|
|
max_seq_len=-1, # Should be unused
|
|
use_pos_emb=False,
|
|
attn_layers=Encoder(
|
|
dim=ch,
|
|
depth=mid_transformer_depth,
|
|
heads=ch//128,
|
|
ff_dropout=dropout,
|
|
attn_dropout=dropout,
|
|
use_rmsnorm=True,
|
|
ff_glu=True,
|
|
rotary_pos_emb=True,
|
|
)
|
|
)
|
|
|
|
self.output_blocks = nn.ModuleList([])
|
|
for level, mult in list(enumerate(channel_mult))[::-1]:
|
|
ich = ch + input_block_chans.pop()
|
|
out_ch = int(model_channels * mult)
|
|
if level != 0:
|
|
self.output_blocks.append(
|
|
TimestepEmbedSequential(Upsample(ich, use_conv=True, dims=dims, out_channels=out_ch, factor=2))
|
|
)
|
|
else:
|
|
self.output_blocks.append(
|
|
TimestepEmbedSequential(conv_nd(dims, ich, out_ch, 3, padding=1))
|
|
)
|
|
ch = out_ch
|
|
ds //= 2
|
|
|
|
self.out = nn.Sequential(
|
|
normalization(ch),
|
|
nn.SiLU(),
|
|
zero_module(conv_nd(dims, model_channels, out_channels, 3, padding=1)),
|
|
)
|
|
|
|
|
|
def forward(self, x, timesteps, tokens=None, conditioning_input=None, lr_input=None, unaligned_input=None):
|
|
"""
|
|
Apply the model to an input batch.
|
|
|
|
:param x: an [N x C x ...] Tensor of inputs.
|
|
:param timesteps: a 1-D batch of timesteps.
|
|
:param tokens: an aligned text input.
|
|
:param conditioning_input: a full-resolution audio clip that is used as a reference to the style you want decoded.
|
|
:param lr_input: for super-sampling models, a guidance audio clip at a lower sampling rate.
|
|
:param unaligned_input: A structural input that is not properly aligned with the output of the diffusion model.
|
|
Can be combined with a conditioning input to produce more robust conditioning.
|
|
:return: an [N x C x ...] Tensor of outputs.
|
|
"""
|
|
assert conditioning_input is not None
|
|
if self.super_sampling_enabled:
|
|
assert lr_input is not None
|
|
if self.training and self.super_sampling_max_noising_factor > 0:
|
|
noising_factor = random.uniform(0,self.super_sampling_max_noising_factor)
|
|
lr_input = torch.randn_like(lr_input) * noising_factor + lr_input
|
|
lr_input = F.interpolate(lr_input, size=(x.shape[-1],), mode='nearest')
|
|
x = torch.cat([x, lr_input], dim=1)
|
|
|
|
if self.enable_unaligned_inputs:
|
|
assert unaligned_input is not None
|
|
unaligned_h = self.unaligned_embedder(unaligned_input).permute(0,2,1)
|
|
unaligned_h = self.unaligned_encoder(unaligned_h).permute(0,2,1)
|
|
|
|
with autocast(x.device.type):
|
|
orig_x_shape = x.shape[-1]
|
|
cm = ceil_multiple(x.shape[-1], 2048)
|
|
if cm != 0:
|
|
pc = (cm-x.shape[-1])/x.shape[-1]
|
|
x = F.pad(x, (0,cm-x.shape[-1]))
|
|
if tokens is not None:
|
|
tokens = F.pad(tokens, (0,int(pc*tokens.shape[-1])))
|
|
|
|
hs = []
|
|
time_emb = self.time_embed(timestep_embedding(timesteps, self.model_channels))
|
|
|
|
cond_emb = self.contextual_embedder(conditioning_input)
|
|
if tokens is not None:
|
|
# Mask out guidance tokens for un-guided diffusion.
|
|
if self.training and self.nil_guidance_fwd_proportion > 0:
|
|
token_mask = torch.rand(tokens.shape, device=tokens.device) < self.nil_guidance_fwd_proportion
|
|
tokens = torch.where(token_mask, self.mask_token_id, tokens)
|
|
code_emb = self.code_embedding(tokens).permute(0,2,1)
|
|
cond_emb = cond_emb.unsqueeze(-1).repeat(1,1,code_emb.shape[-1])
|
|
cond_time_emb = timestep_embedding(torch.zeros_like(timesteps), code_emb.shape[1]) # This was something I was doing (adding timesteps into this computation), but removed on second thought. TODO: completely remove.
|
|
cond_time_emb = cond_time_emb.unsqueeze(-1).repeat(1,1,code_emb.shape[-1])
|
|
code_emb = self.conditioning_conv(torch.cat([cond_emb, code_emb, cond_time_emb], dim=1))
|
|
else:
|
|
code_emb = cond_emb.unsqueeze(-1)
|
|
if self.enable_unaligned_inputs:
|
|
code_emb = self.conditioning_encoder(code_emb, context=unaligned_h)
|
|
else:
|
|
code_emb = self.conditioning_encoder(code_emb)
|
|
|
|
first = True
|
|
time_emb = time_emb.float()
|
|
h = x
|
|
for k, module in enumerate(self.input_blocks):
|
|
if isinstance(module, nn.Conv1d):
|
|
h_tok = F.interpolate(module(code_emb), size=(h.shape[-1]), mode='nearest')
|
|
h = h + h_tok
|
|
else:
|
|
with autocast(x.device.type, enabled=not first):
|
|
# First block has autocast disabled to allow a high precision signal to be properly vectorized.
|
|
h = module(h, time_emb)
|
|
hs.append(h)
|
|
first = False
|
|
h = self.middle_block(h)
|
|
for module in self.output_blocks:
|
|
h = torch.cat([h, hs.pop()], dim=1)
|
|
h = module(h, time_emb)
|
|
|
|
# Last block also has autocast disabled for high-precision outputs.
|
|
h = h.float()
|
|
out = self.out(h)
|
|
return out[:, :, :orig_x_shape]
|
|
|
|
|
|
@register_model
|
|
def register_diffusion_tts8(opt_net, opt):
|
|
return DiffusionTts(**opt_net['kwargs'])
|
|
|
|
|
|
# Test for ~4 second audio clip at 22050Hz
|
|
if __name__ == '__main__':
|
|
clip = torch.randn(2, 1, 32768)
|
|
tok = torch.randint(0,30, (2,388))
|
|
cond = torch.randn(2, 1, 44000)
|
|
ts = torch.LongTensor([600, 600])
|
|
lr = torch.randn(2,1,10000)
|
|
un = torch.randint(0,120, (2,100))
|
|
model = DiffusionTts(128,
|
|
channel_mult=[1,1.5,2, 3, 4, 6, 8],
|
|
token_conditioning_resolutions=[1,4,16,64],
|
|
time_embed_dim_multiplier=4, super_sampling=False,
|
|
enabled_unaligned_inputs=True)
|
|
model(clip, ts, tok, cond, lr, un)
|
|
|