def __init__(self, hparams):
super(Tacotron2, self).__init__()
self.mask_padding = hparams.mask_padding
self.fp16_run = hparams.fp16_run
self.n_mel_channels = hparams.n_mel_channels
self.n_frames_per_step = hparams.n_frames_per_step
self.embedding = nn.Embedding(hparams.n_symbols,
hparams.symbols_embedding_dim)
std = sqrt(2.0 / (hparams.n_symbols + hparams.symbols_embedding_dim))
val = sqrt(3.0) * std # uniform bounds for std
self.embedding.weight.data.uniform_(-val, val)
self.encoder = Encoder(hparams)
self.decoder = Decoder(hparams)
self.postnet = Postnet(hparams)
if hparams.with_gst:
self.gst = GST(hparams)
self.speaker_embedding = nn.Embedding(hparams.n_speakers,
hparams.speaker_embedding_dim)
if hparams.pretrained_speaker:
with open('data/VCTK/speaker-dict.json') as f:
speakers = json.load(f)
embed_paths = glob.glob(
os.path.join(hparams.pretrained_speaker_path, '*.npy'))
embeds = np.zeros((hparams.n_speakers, 1, 512))
for embed_path in embed_paths:
speaker_id = embed_path.split('spker_embed-')[1][:-4]
embed = np.load(embed_path)
embeds[speakers[speaker_id], :, :] = embed
embeddings = torch.tensor(embeds, dtype=torch.float32)
self.register_buffer('pretrained_speaker', embeddings)
self.speaker_linear = nn.Sequential(
nn.Linear(512, hparams.speaker_embedding_dim), nn.ReLU())
def __init__(self, hparams):
super(Tacotron2, self).__init__()
self.mask_padding = hparams.mask_padding
self.fp16_run = hparams.fp16_run
self.n_mel_channels = hparams.n_mel_channels
self.n_frames_per_step = hparams.n_frames_per_step
self.embedding = nn.Embedding(hparams.n_symbols,
hparams.symbols_embedding_dim)
std = sqrt(2.0 / (hparams.n_symbols + hparams.symbols_embedding_dim))
val = sqrt(3.0) * std # uniform bounds for std
self.embedding.weight.data.uniform_(-val, val)
self.encoder = Encoder(hparams)
self.decoder = Decoder(hparams)
self.postnet = Postnet(hparams)
self.gst = GST(hparams)