DL-Art-School/codes/models/audio/tts/unified_voice2.py

589 lines
28 KiB
Python
Raw Normal View History

2022-01-10 05:32:34 +00:00
import torch
import torch.nn as nn
import torch.nn.functional as F
from transformers import GPT2Config, GPT2PreTrainedModel
from transformers.modeling_outputs import CausalLMOutputWithCrossAttentions
2022-04-08 02:11:14 +00:00
from transformers.models.gpt2.modeling_gpt2 import GPT2Attention
from transformers.utils.model_parallel_utils import get_device_map, assert_device_map
2022-01-10 05:32:34 +00:00
from models.arch_util import AttentionBlock
from models.audio.tts.transformer_builders import build_hf_gpt_transformer
2022-04-08 02:11:14 +00:00
from models.lucidrains.x_transformers import RotaryEmbedding, apply_rotary_pos_emb
2022-01-10 05:32:34 +00:00
from trainer.networks import register_model
from utils.util import opt_get
2022-02-04 18:06:17 +00:00
class ResBlock(nn.Module):
"""
Basic residual convolutional block that uses GroupNorm.
"""
def __init__(self, chan):
super().__init__()
self.net = nn.Sequential(
nn.Conv1d(chan, chan, kernel_size=3, padding=1),
nn.GroupNorm(chan//8, chan),
nn.ReLU(),
nn.Conv1d(chan, chan, kernel_size=3, padding=1),
nn.GroupNorm(chan//8, chan)
)
def forward(self, x):
return F.relu(self.net(x) + x)
class GPT2InferenceModel(GPT2PreTrainedModel):
def __init__(self, config, gpt, text_pos_emb, embeddings, norm, linear):
super().__init__(config)
self.transformer = gpt
self.text_pos_embedding = text_pos_emb
self.embeddings = embeddings
self.lm_head = nn.Sequential(norm, linear)
# Model parallel
self.model_parallel = False
self.device_map = None
self.cached_mel_emb = None
def parallelize(self, device_map=None):
self.device_map = (
get_device_map(len(self.transformer.h), range(torch.cuda.device_count()))
if device_map is None
else device_map
)
assert_device_map(self.device_map, len(self.transformer.h))
self.transformer.parallelize(self.device_map)
self.lm_head = self.lm_head.to(self.transformer.first_device)
self.model_parallel = True
def deparallelize(self):
self.transformer.deparallelize()
self.transformer = self.transformer.to("cpu")
self.lm_head = self.lm_head.to("cpu")
self.model_parallel = False
torch.cuda.empty_cache()
def get_output_embeddings(self):
return self.lm_head
def set_output_embeddings(self, new_embeddings):
self.lm_head = new_embeddings
def store_mel_emb(self, mel_emb):
self.cached_mel_emb = mel_emb
def prepare_inputs_for_generation(self, input_ids, past=None, **kwargs):
token_type_ids = kwargs.get("token_type_ids", None)
# only last token for inputs_ids if past is defined in kwargs
if past:
input_ids = input_ids[:, -1].unsqueeze(-1)
if token_type_ids is not None:
token_type_ids = token_type_ids[:, -1].unsqueeze(-1)
attention_mask = kwargs.get("attention_mask", None)
position_ids = kwargs.get("position_ids", None)
if attention_mask is not None and position_ids is None:
# create position_ids on the fly for batch generation
position_ids = attention_mask.long().cumsum(-1) - 1
position_ids.masked_fill_(attention_mask == 0, 1)
if past:
position_ids = position_ids[:, -1].unsqueeze(-1)
else:
position_ids = None
return {
"input_ids": input_ids,
"past_key_values": past,
"use_cache": kwargs.get("use_cache"),
"position_ids": position_ids,
"attention_mask": attention_mask,
"token_type_ids": token_type_ids,
}
def forward(
self,
input_ids=None,
past_key_values=None,
attention_mask=None,
token_type_ids=None,
position_ids=None,
head_mask=None,
inputs_embeds=None,
encoder_hidden_states=None,
encoder_attention_mask=None,
labels=None,
use_cache=None,
output_attentions=None,
output_hidden_states=None,
return_dict=None,
):
assert self.cached_mel_emb is not None
assert inputs_embeds is None # Not supported by this inference model.
assert labels is None # Training not supported by this inference model.
return_dict = return_dict if return_dict is not None else self.config.use_return_dict
# Create embedding
mel_len = self.cached_mel_emb.shape[1]
if input_ids.shape[1] != 1:
text_inputs = input_ids[:, mel_len:]
text_emb = self.embeddings(text_inputs)
text_emb = text_emb + self.text_pos_embedding(text_emb)
if self.cached_mel_emb.shape[0] != text_emb.shape[0]:
mel_emb = self.cached_mel_emb.repeat_interleave(text_emb.shape[0]//self.cached_mel_emb.shape[0], 0)
else:
mel_emb = self.cached_mel_emb
emb = torch.cat([mel_emb, text_emb], dim=1)
else:
emb = self.embeddings(input_ids)
emb = emb + self.text_pos_embedding.get_fixed_embedding(attention_mask.shape[1]-mel_len, attention_mask.device)
transformer_outputs = self.transformer(
inputs_embeds=emb,
past_key_values=past_key_values,
attention_mask=attention_mask,
token_type_ids=token_type_ids,
position_ids=position_ids,
head_mask=head_mask,
encoder_hidden_states=encoder_hidden_states,
encoder_attention_mask=encoder_attention_mask,
use_cache=use_cache,
output_attentions=output_attentions,
output_hidden_states=output_hidden_states,
return_dict=return_dict,
)
hidden_states = transformer_outputs[0]
# Set device for model parallelism
if self.model_parallel:
torch.cuda.set_device(self.transformer.first_device)
hidden_states = hidden_states.to(self.lm_head.weight.device)
lm_logits = self.lm_head(hidden_states)
if not return_dict:
return (lm_logits,) + transformer_outputs[1:]
return CausalLMOutputWithCrossAttentions(
loss=None,
logits=lm_logits,
past_key_values=transformer_outputs.past_key_values,
hidden_states=transformer_outputs.hidden_states,
attentions=transformer_outputs.attentions,
cross_attentions=transformer_outputs.cross_attentions,
)
@staticmethod
def _reorder_cache(past, beam_idx):
"""
This function is used to re-order the :obj:`past_key_values` cache if
:meth:`~transformers.PreTrainedModel.beam_search` or :meth:`~transformers.PreTrainedModel.beam_sample` is
called. This is required to match :obj:`past_key_values` with the correct beam_idx at every generation step.
"""
return tuple(
tuple(past_state.index_select(0, beam_idx.to(past_state.device)) for past_state in layer_past)
for layer_past in past
)
2022-01-10 05:32:34 +00:00
class ConditioningEncoder(nn.Module):
def __init__(self,
spec_dim,
embedding_dim,
attn_blocks=6,
num_attn_heads=4,
do_checkpointing=False,
mean=False):
2022-01-10 05:32:34 +00:00
super().__init__()
attn = []
self.init = nn.Conv1d(spec_dim, embedding_dim, kernel_size=1)
for a in range(attn_blocks):
attn.append(AttentionBlock(embedding_dim, num_attn_heads, do_checkpoint=do_checkpointing))
self.attn = nn.Sequential(*attn)
self.dim = embedding_dim
self.do_checkpointing = do_checkpointing
self.mean = mean
2022-01-10 05:32:34 +00:00
def forward(self, x):
h = self.init(x)
h = self.attn(h)
if self.mean:
return h.mean(dim=2)
else:
return h[:, :, 0]
2022-01-10 05:32:34 +00:00
class MelEncoder(nn.Module):
def __init__(self, channels, mel_channels=80, resblocks_per_reduction=2):
super().__init__()
self.channels = channels
self.encoder = nn.Sequential(nn.Conv1d(mel_channels, channels//4, kernel_size=3, padding=1),
nn.Sequential(*[ResBlock(channels//4) for _ in range(resblocks_per_reduction)]),
nn.Conv1d(channels//4, channels//2, kernel_size=3, stride=2, padding=1),
nn.GroupNorm(channels//16, channels//2),
nn.ReLU(),
nn.Sequential(*[ResBlock(channels//2) for _ in range(resblocks_per_reduction)]),
nn.Conv1d(channels//2, channels, kernel_size=3, stride=2, padding=1),
nn.GroupNorm(channels//8, channels),
nn.ReLU(),
nn.Sequential(*[ResBlock(channels) for _ in range(resblocks_per_reduction)]),
)
self.reduction = 4
def forward(self, x):
for e in self.encoder:
x = e(x)
return x.permute(0,2,1)
class UnifiedVoice(nn.Module):
def __init__(self, layers=8, model_dim=512, heads=8, max_text_tokens=120, max_mel_tokens=250, max_conditioning_inputs=1,
2022-01-10 23:17:31 +00:00
mel_length_compression=1024, number_text_tokens=256,
2022-01-10 05:32:34 +00:00
start_text_token=255, stop_text_token=0, number_mel_codes=8194, start_mel_token=8192,
stop_mel_token=8193, train_solo_embeddings=False, use_mel_codes_as_input=True,
checkpointing=True, average_conditioning_embeddings=False, freeze_everything_but_position_embeddings=False):
2022-01-10 05:32:34 +00:00
"""
Args:
layers: Number of layers in transformer stack.
model_dim: Operating dimensions of the transformer
heads: Number of transformer heads. Must be divisible by model_dim. Recommend model_dim//64
max_text_tokens: Maximum number of text tokens that will be encountered by model.
max_mel_tokens: Maximum number of MEL tokens that will be encountered by model.
max_conditioning_inputs: Maximum number of conditioning inputs provided to the model. If (1), conditioning input can be of format (b,80,s), otherwise (b,n,80,s).
mel_length_compression: The factor between <number_input_samples> and <mel_tokens>. Used to compute MEL code padding given wav input length.
number_text_tokens:
start_text_token:
stop_text_token:
number_mel_codes:
start_mel_token:
stop_mel_token:
train_solo_embeddings:
use_mel_codes_as_input:
checkpointing:
average_conditioning_embeddings: Whether or not conditioning embeddings should be averaged, instead of fed piecewise into the model.
2022-01-10 05:32:34 +00:00
"""
super().__init__()
self.number_text_tokens = number_text_tokens
self.start_text_token = start_text_token
self.stop_text_token = stop_text_token
self.number_mel_codes = number_mel_codes
self.start_mel_token = start_mel_token
self.stop_mel_token = stop_mel_token
self.layers = layers
self.heads = heads
self.max_conditioning_inputs = max_conditioning_inputs
2022-04-08 02:11:14 +00:00
self.max_mel_tokens = -1 if max_mel_tokens == -1 else max_mel_tokens+2+self.max_conditioning_inputs
self.max_text_tokens = -1 if max_text_tokens == -1 else max_text_tokens+2
2022-01-10 05:32:34 +00:00
self.model_dim = model_dim
self.mel_length_compression = mel_length_compression
self.conditioning_encoder = ConditioningEncoder(80, model_dim, num_attn_heads=heads)
self.average_conditioning_embeddings = average_conditioning_embeddings
2022-01-10 05:32:34 +00:00
self.text_embedding = nn.Embedding(self.number_text_tokens, model_dim)
if use_mel_codes_as_input:
self.mel_embedding = nn.Embedding(self.number_mel_codes, model_dim)
else:
self.mel_embedding = MelEncoder(model_dim, resblocks_per_reduction=1)
self.gpt, self.mel_pos_embedding, self.text_pos_embedding, self.mel_layer_pos_embedding, self.text_layer_pos_embedding = \
2022-04-08 02:11:14 +00:00
build_hf_gpt_transformer(layers, model_dim, heads, self.max_mel_tokens, self.max_text_tokens, checkpointing)
2022-01-10 05:32:34 +00:00
if train_solo_embeddings:
self.mel_solo_embedding = nn.Parameter(torch.randn(1, 1, model_dim) * .02, requires_grad=True)
self.text_solo_embedding = nn.Parameter(torch.randn(1, 1, model_dim) * .02, requires_grad=True)
else:
self.mel_solo_embedding = 0
self.text_solo_embedding = 0
self.final_norm = nn.LayerNorm(model_dim)
self.text_head = nn.Linear(model_dim, self.number_text_tokens)
self.mel_head = nn.Linear(model_dim, self.number_mel_codes)
# Initialize the embeddings per the GPT-2 scheme
embeddings = [self.text_embedding]
2022-01-10 05:32:34 +00:00
if use_mel_codes_as_input:
embeddings.append(self.mel_embedding)
for module in embeddings:
2022-01-10 05:32:34 +00:00
module.weight.data.normal_(mean=0.0, std=.02)
if freeze_everything_but_position_embeddings:
for p in self.parameters():
p.requires_grad = False
p.DO_NOT_TRAIN = True
for m in [self.mel_pos_embedding, self.text_pos_embedding]:
for p in m.parameters():
del p.DO_NOT_TRAIN
p.requires_grad = True
2022-03-06 03:14:36 +00:00
def get_grad_norm_parameter_groups(self):
return {
'conditioning_encoder': list(self.conditioning_encoder.parameters()),
'gpt': list(self.gpt.parameters()),
'heads': list(self.text_head.parameters()) + list(self.mel_head.parameters()),
}
2022-01-10 05:32:34 +00:00
def build_aligned_inputs_and_targets(self, input, start_token, stop_token):
inp = F.pad(input, (1,0), value=start_token)
tar = F.pad(input, (0,1), value=stop_token)
return inp, tar
def set_mel_padding(self, mel_input_tokens, wav_lengths):
"""
Given mel tokens that are derived from a padded audio clip and the actual lengths of each batch element in
that audio clip, reformats the tokens with STOP_MEL_TOKEN in place of the zero padding. This is required
preformatting to create a working TTS model.
"""
# Set padding areas within MEL (currently it is coded with the MEL code for <zero>).
mel_lengths = wav_lengths // self.mel_length_compression
for b in range(len(mel_lengths)):
actual_end = mel_lengths[b] + 1 # Due to the convolutional nature of how these tokens are generated, it would be best if the model predicts a token past the actual last token.
if actual_end < mel_input_tokens.shape[-1]:
mel_input_tokens[b, actual_end:] = self.stop_mel_token
return mel_input_tokens
2022-03-06 03:14:36 +00:00
def get_logits(self, speech_conditioning_inputs, first_inputs, first_head, second_inputs=None, second_head=None, get_attns=False, return_latent=False):
2022-01-10 05:32:34 +00:00
if second_inputs is not None:
2022-01-10 23:17:31 +00:00
emb = torch.cat([speech_conditioning_inputs, first_inputs, second_inputs], dim=1)
2022-01-10 05:32:34 +00:00
else:
2022-01-10 23:17:31 +00:00
emb = torch.cat([speech_conditioning_inputs, first_inputs], dim=1)
2022-01-10 05:32:34 +00:00
gpt_out = self.gpt(inputs_embeds=emb, return_dict=True, output_attentions=get_attns)
if get_attns:
return gpt_out.attentions
enc = gpt_out.last_hidden_state[:, 1:] # The first logit is tied to the speech_conditioning_input
enc = self.final_norm(enc)
2022-03-06 03:14:36 +00:00
if return_latent:
return enc[:, speech_conditioning_inputs.shape[1]:speech_conditioning_inputs.shape[1]+first_inputs.shape[1]], enc[:, -second_inputs.shape[1]:]
2022-03-06 03:14:36 +00:00
2022-01-10 05:32:34 +00:00
first_logits = enc[:, :first_inputs.shape[1]]
first_logits = first_head(first_logits)
first_logits = first_logits.permute(0,2,1)
if second_inputs is not None:
second_logits = enc[:, -second_inputs.shape[1]:]
second_logits = second_head(second_logits)
second_logits = second_logits.permute(0,2,1)
return first_logits, second_logits
else:
return first_logits
2022-03-06 03:14:36 +00:00
def forward(self, speech_conditioning_input, text_inputs, text_lengths, mel_codes, wav_lengths, text_first=True, raw_mels=None, return_attentions=False,
return_latent=False, clip_inputs=True):
2022-01-10 05:32:34 +00:00
"""
Forward pass that uses both text and voice in either text conditioning mode or voice conditioning mode
(actuated by `text_first`).
speech_conditioning_input: MEL float tensor, (b,80,s)
text_inputs: long tensor, (b,t)
text_lengths: long tensor, (b,)
mel_inputs: long tensor, (b,m)
wav_lengths: long tensor, (b,)
raw_mels: MEL float tensor (b,80,s)
2022-03-06 03:14:36 +00:00
If return_attentions is specified, only logits are returned.
If return_latent is specified, loss & logits are not computed or returned. Only the predicted latents are returned.
If clip_inputs is True, the inputs will be clipped to the smallest input size across each input modality.
2022-01-10 05:32:34 +00:00
"""
if clip_inputs:
# This model will receive micro-batches with a ton of padding for both the text and MELs. Ameliorate this by
# chopping the inputs by the maximum actual length.
max_text_len = text_lengths.max()
text_inputs = text_inputs[:, :max_text_len]
max_mel_len = wav_lengths.max() // self.mel_length_compression
mel_codes = mel_codes[:, :max_mel_len]
if raw_mels is not None:
raw_mels = raw_mels[:, :, :max_mel_len*4]
2022-01-10 05:32:34 +00:00
mel_codes = self.set_mel_padding(mel_codes, wav_lengths)
text_inputs = F.pad(text_inputs, (0,1), value=self.stop_text_token)
mel_codes = F.pad(mel_codes, (0,1), value=self.stop_mel_token)
2022-01-10 05:32:34 +00:00
2022-01-10 23:17:31 +00:00
speech_conditioning_input = speech_conditioning_input.unsqueeze(1) if len(speech_conditioning_input.shape) == 3 else speech_conditioning_input
conds = []
for j in range(speech_conditioning_input.shape[1]):
conds.append(self.conditioning_encoder(speech_conditioning_input[:, j]))
conds = torch.stack(conds, dim=1)
if self.average_conditioning_embeddings:
conds = conds.mean(dim=1).unsqueeze(1)
2022-01-10 05:32:34 +00:00
text_inputs, text_targets = self.build_aligned_inputs_and_targets(text_inputs, self.start_text_token, self.stop_text_token)
text_emb = self.text_embedding(text_inputs) + self.text_pos_embedding(text_inputs)
2022-01-10 05:32:34 +00:00
mel_codes, mel_targets = self.build_aligned_inputs_and_targets(mel_codes, self.start_mel_token, self.stop_mel_token)
if raw_mels is not None:
mel_inp = F.pad(raw_mels, (0, 8))
else:
mel_inp = mel_codes
mel_emb = self.mel_embedding(mel_inp)
mel_emb = mel_emb + self.mel_pos_embedding(mel_codes)
2022-03-06 03:14:36 +00:00
2022-01-10 05:32:34 +00:00
if text_first:
2022-03-06 03:14:36 +00:00
text_logits, mel_logits = self.get_logits(conds, text_emb, self.text_head, mel_emb, self.mel_head, get_attns=return_attentions, return_latent=return_latent)
if return_latent:
return mel_logits[:, :-2] # Despite the name, these are not logits. Strip off the two tokens added by this forward pass.
2022-01-10 05:32:34 +00:00
else:
2022-03-06 03:14:36 +00:00
mel_logits, text_logits = self.get_logits(conds, mel_emb, self.mel_head, text_emb, self.text_head, get_attns=return_attentions, return_latent=return_latent)
if return_latent:
return text_logits[:, :-2] # Despite the name, these are not logits. Strip off the two tokens added by this forward pass.
2022-01-10 05:32:34 +00:00
if return_attentions:
return mel_logits
loss_text = F.cross_entropy(text_logits, text_targets.long())
loss_mel = F.cross_entropy(mel_logits, mel_targets.long())
return loss_text.mean(), loss_mel.mean(), mel_logits
def text_forward(self, speech_conditioning_input, text_inputs, text_lengths):
"""
Performs autoregressive modeling on only text. Still requires a speech_conditioning_input due to the way the
model inputs are formatted. Just provide any audio clip (arguably, zeros could be provided).
"""
# This model will receive micro-batches with a ton of padding for both the text and MELs. Ameliorate this by
# chopping the inputs by the maximum actual length.
max_text_len = text_lengths.max()
text_inputs = F.pad(text_inputs[:, :max_text_len], (0,1), value=self.stop_text_token)
2022-01-10 23:17:31 +00:00
speech_conditioning_input = speech_conditioning_input.unsqueeze(1) if len(speech_conditioning_input.shape) == 3 else speech_conditioning_input
conds = []
for j in range(speech_conditioning_input.shape[1]):
conds.append(self.conditioning_encoder(speech_conditioning_input[:, j]))
conds = torch.stack(conds, dim=1)
2022-03-21 20:44:00 +00:00
if self.average_conditioning_embeddings:
conds = conds.mean(dim=1).unsqueeze(1)
2022-01-10 05:32:34 +00:00
text_inputs, text_targets = self.build_aligned_inputs_and_targets(text_inputs, self.start_text_token, self.stop_text_token)
text_emb = self.text_embedding(text_inputs) + self.text_pos_embedding(text_inputs) + self.text_solo_embedding
2022-01-10 23:17:31 +00:00
text_logits = self.get_logits(conds, text_emb, self.text_head)
2022-01-10 05:32:34 +00:00
loss_text = F.cross_entropy(text_logits, text_targets.long())
return loss_text.mean()
def speech_forward(self, speech_conditioning_input, mel_codes, wav_lengths, raw_mels=None):
"""
Performs autoregressive modeling on only speech data.
"""
assert self.max_mel_tokens >= mel_codes.shape[1], f'{mel_codes.shape[1]}'
# This model will receive micro-batches with a ton of padding for both the text and MELs. Ameliorate this by
# chopping the inputs by the maximum actual length.
max_mel_len = wav_lengths.max() // self.mel_length_compression
mel_codes = F.pad(mel_codes[:, :max_mel_len], (0,1), value=self.stop_mel_token)
mel_codes = self.set_mel_padding(mel_codes, wav_lengths)
if raw_mels is not None:
raw_mels = raw_mels[:, :, :max_mel_len*4]
2022-01-10 23:17:31 +00:00
speech_conditioning_input = speech_conditioning_input.unsqueeze(1) if len(speech_conditioning_input.shape) == 3 else speech_conditioning_input
conds = []
for j in range(speech_conditioning_input.shape[1]):
conds.append(self.conditioning_encoder(speech_conditioning_input[:, j]))
conds = torch.stack(conds, dim=1)
2022-03-21 20:44:00 +00:00
if self.average_conditioning_embeddings:
conds = conds.mean(dim=1).unsqueeze(1)
2022-01-10 05:32:34 +00:00
mel_codes, mel_targets = self.build_aligned_inputs_and_targets(mel_codes, self.start_mel_token, self.stop_mel_token)
if raw_mels is not None:
mel_inp = F.pad(raw_mels, (0, 4))
else:
mel_inp = mel_codes
mel_emb = self.mel_embedding(mel_inp)
mel_emb = mel_emb + self.mel_pos_embedding(mel_codes) + self.mel_solo_embedding
2022-01-10 23:17:31 +00:00
mel_logits = self.get_logits(conds, mel_emb, self.mel_head)
2022-01-10 05:32:34 +00:00
loss_mel = F.cross_entropy(mel_logits, mel_targets.long())
return loss_mel.mean()
def inference_speech(self, speech_conditioning_input, text_inputs, return_attentions=False, **hf_generate_kwargs):
2022-04-08 02:11:14 +00:00
if self.max_mel_tokens == -1: # Assume if this is the case, max_mel_tokens=-1 also
seq_length = 2002 # Arbitrary default.
else:
seq_length = self.max_mel_tokens + self.max_text_tokens + 2
2022-01-10 05:32:34 +00:00
if not hasattr(self, 'inference_model'):
# TODO: Decouple gpt_config from this inference model.
gpt_config = GPT2Config(vocab_size=self.max_mel_tokens,
n_positions=seq_length,
n_ctx=seq_length,
2022-01-10 05:32:34 +00:00
n_embd=self.model_dim,
n_layer=self.layers,
n_head=self.heads,
gradient_checkpointing=False,
use_cache=True)
self.inference_model = GPT2InferenceModel(gpt_config, self.gpt, self.mel_pos_embedding, self.mel_embedding, self.final_norm, self.mel_head)
self.gpt.wte = self.mel_embedding
2022-01-10 05:32:34 +00:00
text_inputs = F.pad(text_inputs, (0, 1), value=self.stop_text_token)
text_inputs, text_targets = self.build_aligned_inputs_and_targets(text_inputs, self.start_text_token, self.stop_text_token)
text_emb = self.text_embedding(text_inputs) + self.text_pos_embedding(text_inputs)
2022-01-10 05:32:34 +00:00
2022-01-10 23:17:31 +00:00
speech_conditioning_input = speech_conditioning_input.unsqueeze(1) if len(speech_conditioning_input.shape) == 3 else speech_conditioning_input
conds = []
for j in range(speech_conditioning_input.shape[1]):
conds.append(self.conditioning_encoder(speech_conditioning_input[:, j]))
conds = torch.stack(conds, dim=1)
2022-03-21 20:44:00 +00:00
if self.average_conditioning_embeddings:
conds = conds.mean(dim=1).unsqueeze(1)
2022-01-10 05:32:34 +00:00
2022-01-10 23:17:31 +00:00
emb = torch.cat([conds, text_emb], dim=1)
2022-01-10 05:32:34 +00:00
self.inference_model.store_mel_emb(emb)
fake_inputs = torch.full((emb.shape[0], conds.shape[1]+emb.shape[1],), fill_value=1, dtype=torch.long, device=text_inputs.device)
2022-01-10 05:32:34 +00:00
fake_inputs[:,-1] = self.start_mel_token
gen = self.inference_model.generate(fake_inputs, bos_token_id=self.start_mel_token, pad_token_id=self.stop_mel_token, eos_token_id=self.stop_mel_token,
max_length=seq_length, output_attentions=return_attentions, return_dict_in_generate=True, **hf_generate_kwargs)
if return_attentions:
return gen.sequences[:, fake_inputs.shape[1]:], gen.attentions
else:
return gen.sequences[:, fake_inputs.shape[1]:]
2022-02-20 03:37:35 +00:00
# Turns the (utterly insane) output of HF.generate() into a far more sane output:
# [tensors(B,H,S,S)]. Outer=layers, B=batch,H=head,S=sequence
def make_hf_generate_attentions_sane(self, attentions):
layers = [[] for _ in range(len(attentions[0]))]
full_attention_size = attentions[-1][0].shape[-1]
for i, gen in enumerate(attentions):
for j, lyr in enumerate(gen):
layers[j].append(F.pad(lyr, (0, full_attention_size - lyr.shape[-1])))
catted = []
for lyr in layers:
catted.append(torch.cat(lyr, dim=2))
return catted
def convert_attentions_to_aligned_codes(self, text, attentions, codes, num_conds):
2022-02-20 03:37:35 +00:00
"""
This was an attempt to make some sense out of the attention matrix retrieved from the unified_voice model. Unfortunately, I can't use it for aligning text & voice.
"""
text_padding = num_conds+2
num_text = text.shape[-1]
2022-02-20 03:37:35 +00:00
num_context = num_text + text_padding
assert num_context + 1 == attentions[0][0].shape[-1]
attentions = self.make_hf_generate_attentions_sane(attentions)
results = [torch.empty_like(codes) for _ in range(len(attentions))]
for l, layer in enumerate(attentions):
dec_context = layer[:, :, num_context:, :]
# Mask out everything that isn't text (including the start token, which gets a LOT of attention)
dec_context[:,:,:,:text_padding+1] = 0
dec_context[:,:,:,num_context:] = 0
for h in range(dec_context.shape[1]):
dec_context_indices = torch.argmax(dec_context[0,h], dim=-1)
print(f'layer_{l};head_{h}: ' + str(dec_context_indices))
for t, att_tok in enumerate(attentions):
combined_attention_weights = torch.zeros((codes.shape[0], num_text), device=codes.device)
for lyr in att_tok:
token_to_text_attentions = lyr[:, :, -1, text_padding:(text_padding + num_text)].sum(dim=1)
combined_attention_weights = combined_attention_weights + token_to_text_attentions
break
most_attended_text_token = combined_attention_weights.argmax(dim=-1)
results[:, t] = most_attended_text_token
eos_token_mask = (codes != self.stop_mel_token)
return results * eos_token_mask
2022-01-10 05:32:34 +00:00
@register_model
def register_unified_voice2(opt_net, opt):
return UnifiedVoice(**opt_get(opt_net, ['kwargs'], {}))
if __name__ == '__main__':
gpt = UnifiedVoice(model_dim=256, heads=4, train_solo_embeddings=True, use_mel_codes_as_input=True, max_conditioning_inputs=4, freeze_everything_but_position_embeddings=True)
2022-01-10 23:17:31 +00:00
l = gpt(torch.randn(2, 3, 80, 800),
torch.randint(high=256, size=(2,120)),
2022-01-10 23:17:31 +00:00
torch.tensor([32, 120]),
2022-01-10 05:32:34 +00:00
torch.randint(high=8192, size=(2,250)),
2022-01-10 23:17:31 +00:00
torch.tensor([250*256,195*256]))
2022-04-08 02:11:14 +00:00
#gpt.text_forward(torch.randn(2,80,800), torch.randint(high=50, size=(2,80)), torch.tensor([32, 80]))