StarMirror/MockingBird
1
0
Fork 0
mirror of https://github.com/babysor/MockingBird.git synced 2024-03-22 13:11:31 +08:00
Code Issues Projects Releases Wiki Activity

Add max steps control in toolbox

Browse Source
This commit is contained in:
babysor00 2021-11-06 13:27:11 +08:00
parent c396792b22
commit 80aaf32164
4 changed files with 17 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
synthesizer/inference.py
View File

@ -70,7 +70,7 @@ class Synthesizer:
def synthesize_spectrograms(self, texts: List[str], def synthesize_spectrograms(self, texts: List[str],
embeddings: Union[np.ndarray, List[np.ndarray]], embeddings: Union[np.ndarray, List[np.ndarray]],
return_alignments=False, style_idx=0, min_stop_token=5): return_alignments=False, style_idx=0, min_stop_token=5, steps=2000):
""" """
Synthesizes mel spectrograms from texts and speaker embeddings. Synthesizes mel spectrograms from texts and speaker embeddings.
@ -125,7 +125,7 @@ class Synthesizer:
speaker_embeddings = torch.tensor(speaker_embeds).float().to(self.device) speaker_embeddings = torch.tensor(speaker_embeds).float().to(self.device)
# Inference # Inference
_, mels, alignments = self._model.generate(chars, speaker_embeddings, style_idx=style_idx, min_stop_token=min_stop_token) _, mels, alignments = self._model.generate(chars, speaker_embeddings, style_idx=style_idx, min_stop_token=min_stop_token, steps=steps)
mels = mels.detach().cpu().numpy() mels = mels.detach().cpu().numpy()
for m in mels: for m in mels:
# Trim silence from end of each spectrogram # Trim silence from end of each spectrogram

2
synthesizer/models/tacotron.py
View File

@ -420,7 +420,7 @@ class Tacotron(nn.Module):
return mel_outputs, linear, attn_scores, stop_outputs return mel_outputs, linear, attn_scores, stop_outputs
def generate(self, x, speaker_embedding=None, steps=200, style_idx=0, min_stop_token=5): def generate(self, x, speaker_embedding=None, steps=2000, style_idx=0, min_stop_token=5):
self.eval() self.eval()
device = next(self.parameters()).device # use same device as parameters device = next(self.parameters()).device # use same device as parameters

2
toolbox/__init__.py
View File

@ -235,7 +235,7 @@ class Toolbox:
embed = self.ui.selected_utterance.embed embed = self.ui.selected_utterance.embed
embeds = [embed] * len(texts) embeds = [embed] * len(texts)
min_token = int(self.ui.token_slider.value()) min_token = int(self.ui.token_slider.value())
specs = self.synthesizer.synthesize_spectrograms(texts, embeds, style_idx=int(self.ui.style_slider.value()), min_stop_token=min_token) specs = self.synthesizer.synthesize_spectrograms(texts, embeds, style_idx=int(self.ui.style_slider.value()), min_stop_token=min_token, steps=int(self.ui.length_slider.value())*200)
breaks = [spec.shape[1] for spec in specs] breaks = [spec.shape[1] for spec in specs]
spec = np.concatenate(specs, axis=1) spec = np.concatenate(specs, axis=1)

13
toolbox/ui.py
View File

@ -618,6 +618,19 @@ class UI(QDialog):
layout_seed.addWidget(self.token_value_label, 2, 1) layout_seed.addWidget(self.token_value_label, 2, 1)
layout_seed.addWidget(self.token_slider, 2, 3) layout_seed.addWidget(self.token_slider, 2, 3)
self.length_slider = QSlider(Qt.Horizontal)
self.length_slider.setTickInterval(1)
self.length_slider.setFocusPolicy(Qt.NoFocus)
self.length_slider.setSingleStep(1)
self.length_slider.setRange(1, 10)
self.length_value_label = QLabel("2")
self.length_slider.setValue(2)
layout_seed.addWidget(QLabel("MaxLength(最大句长):"), 3, 0)
self.length_slider.valueChanged.connect(lambda s: self.length_value_label.setNum(s))
layout_seed.addWidget(self.length_value_label, 3, 1)
layout_seed.addWidget(self.length_slider, 3, 3)
gen_layout.addLayout(layout_seed) gen_layout.addLayout(layout_seed)
self.loading_bar = QProgressBar() self.loading_bar = QProgressBar()