Use speaker embedding anyway even with default style

This commit is contained in:
babysor00 2021-11-07 21:48:15 +08:00
parent 80aaf32164
commit 3674d8b5c6
2 changed files with 55 additions and 14 deletions

View File

@ -6,15 +6,22 @@ from synthesizer.gst_hyperparameters import GSTHyperparameters as hp
class GlobalStyleToken(nn.Module): class GlobalStyleToken(nn.Module):
"""
def __init__(self): inputs: style mel spectrograms [batch_size, num_spec_frames, num_mel]
speaker_embedding: speaker mel spectrograms [batch_size, num_spec_frames, num_mel]
outputs: [batch_size, embedding_dim]
"""
def __init__(self, speaker_embedding_dim=None):
super().__init__() super().__init__()
self.encoder = ReferenceEncoder() self.encoder = ReferenceEncoder()
self.stl = STL() self.stl = STL(speaker_embedding_dim)
def forward(self, inputs): def forward(self, inputs, speaker_embedding=None):
enc_out = self.encoder(inputs) enc_out = self.encoder(inputs)
# concat speaker_embedding according to https://github.com/mozilla/TTS/blob/master/TTS/tts/layers/gst_layers.py
if speaker_embedding is not None:
enc_out = torch.cat([enc_out, speaker_embedding], dim=-1)
style_embed = self.stl(enc_out) style_embed = self.stl(enc_out)
return style_embed return style_embed
@ -73,13 +80,15 @@ class STL(nn.Module):
inputs --- [N, E//2] inputs --- [N, E//2]
''' '''
def __init__(self): def __init__(self, speaker_embedding_dim=None):
super().__init__() super().__init__()
self.embed = nn.Parameter(torch.FloatTensor(hp.token_num, hp.E // hp.num_heads)) self.embed = nn.Parameter(torch.FloatTensor(hp.token_num, hp.E // hp.num_heads))
d_q = hp.E // 2 d_q = hp.E // 2
d_k = hp.E // hp.num_heads d_k = hp.E // hp.num_heads
# self.attention = MultiHeadAttention(hp.num_heads, d_model, d_q, d_v) # self.attention = MultiHeadAttention(hp.num_heads, d_model, d_q, d_v)
if speaker_embedding_dim:
d_q += speaker_embedding_dim
self.attention = MultiHeadAttention(query_dim=d_q, key_dim=d_k, num_units=hp.E, num_heads=hp.num_heads) self.attention = MultiHeadAttention(query_dim=d_q, key_dim=d_k, num_units=hp.E, num_heads=hp.num_heads)
init.normal_(self.embed, mean=0, std=0.5) init.normal_(self.embed, mean=0, std=0.5)

View File

@ -338,7 +338,7 @@ class Tacotron(nn.Module):
self.encoder = Encoder(embed_dims, num_chars, encoder_dims, self.encoder = Encoder(embed_dims, num_chars, encoder_dims,
encoder_K, num_highways, dropout) encoder_K, num_highways, dropout)
self.encoder_proj = nn.Linear(encoder_dims + speaker_embedding_size + gst_hp.E, decoder_dims, bias=False) self.encoder_proj = nn.Linear(encoder_dims + speaker_embedding_size + gst_hp.E, decoder_dims, bias=False)
self.gst = GlobalStyleToken() self.gst = GlobalStyleToken(speaker_embedding_size)
self.decoder = Decoder(n_mels, encoder_dims, decoder_dims, lstm_dims, self.decoder = Decoder(n_mels, encoder_dims, decoder_dims, lstm_dims,
dropout, speaker_embedding_size) dropout, speaker_embedding_size)
self.postnet = CBHG(postnet_K, n_mels, postnet_dims, self.postnet = CBHG(postnet_K, n_mels, postnet_dims,
@ -359,6 +359,34 @@ class Tacotron(nn.Module):
def r(self, value): def r(self, value):
self.decoder.r = self.decoder.r.new_tensor(value, requires_grad=False) self.decoder.r = self.decoder.r.new_tensor(value, requires_grad=False)
def compute_gst(self, inputs, style_input, speaker_embedding=None):
""" Compute global style token """
device = inputs.device
if isinstance(style_input, dict):
query = torch.zeros(1, 1, self.gst_embedding_dim//2).to(device)
if speaker_embedding is not None:
query = torch.cat([query, speaker_embedding.reshape(1, 1, -1)], dim=-1)
_GST = torch.tanh(self.gst_layer.style_token_layer.style_tokens)
gst_outputs = torch.zeros(1, 1, self.gst_embedding_dim).to(device)
for k_token, v_amplifier in style_input.items():
key = _GST[int(k_token)].unsqueeze(0).expand(1, -1, -1)
gst_outputs_att = self.gst_layer.style_token_layer.attention(query, key)
gst_outputs = gst_outputs + gst_outputs_att * v_amplifier
elif style_input is None:
gst_outputs = torch.zeros(1, 1, self.gst_embedding_dim).to(device)
else:
gst_outputs = self.gst_layer(style_input, speaker_embedding) # pylint: disable=not-callable
inputs = self._concat_speaker_embedding(inputs, gst_outputs)
return inputs
@staticmethod
def _concat_speaker_embedding(outputs, speaker_embeddings):
speaker_embeddings_ = speaker_embeddings.expand(
outputs.size(0), outputs.size(1), -1)
outputs = torch.cat([outputs, speaker_embeddings_], dim=-1)
return outputs
def forward(self, texts, mels, speaker_embedding): def forward(self, texts, mels, speaker_embedding):
device = next(self.parameters()).device # use same device as parameters device = next(self.parameters()).device # use same device as parameters
@ -387,9 +415,10 @@ class Tacotron(nn.Module):
encoder_seq = self.encoder(texts, speaker_embedding) encoder_seq = self.encoder(texts, speaker_embedding)
# put after encoder # put after encoder
if self.gst is not None: if self.gst is not None:
style_embed = self.gst(speaker_embedding) style_embed = self.gst(speaker_embedding, speaker_embedding) # for training, speaker embedding can represent both style inputs and referenced
style_embed = style_embed.expand_as(encoder_seq) # style_embed = style_embed.expand_as(encoder_seq)
encoder_seq = torch.cat((encoder_seq, style_embed), 2) # encoder_seq = torch.cat((encoder_seq, style_embed), 2)
encoder_seq = self._concat_speaker_embedding(encoder_seq, style_embed)
encoder_seq_proj = self.encoder_proj(encoder_seq) encoder_seq_proj = self.encoder_proj(encoder_seq)
# Need a couple of lists for outputs # Need a couple of lists for outputs
@ -454,11 +483,14 @@ class Tacotron(nn.Module):
gst_embed = np.tile(gst_embed, (1, 8)) gst_embed = np.tile(gst_embed, (1, 8))
scale = np.zeros(512) scale = np.zeros(512)
scale[:] = 0.3 scale[:] = 0.3
speaker_embedding = (gst_embed[style_idx] * scale).astype(np.float32) speaker_embedding_style = (gst_embed[style_idx] * scale).astype(np.float32)
speaker_embedding = torch.from_numpy(np.tile(speaker_embedding, (x.shape[0], 1))).to(device) speaker_embedding_style = torch.from_numpy(np.tile(speaker_embedding_style, (x.shape[0], 1))).to(device)
style_embed = self.gst(speaker_embedding) else:
style_embed = style_embed.expand_as(encoder_seq) speaker_embedding_style = torch.zeros(2, 1, self.speaker_embedding_size).to(device)
encoder_seq = torch.cat((encoder_seq, style_embed), 2) style_embed = self.gst(speaker_embedding_style, speaker_embedding)
encoder_seq = self._concat_speaker_embedding(encoder_seq, style_embed)
# style_embed = style_embed.expand_as(encoder_seq)
# encoder_seq = torch.cat((encoder_seq, style_embed), 2)
encoder_seq_proj = self.encoder_proj(encoder_seq) encoder_seq_proj = self.encoder_proj(encoder_seq)
# Need a couple of lists for outputs # Need a couple of lists for outputs