mirror of
https://github.com/wassname/TTS.git
synced 2026-10-29 12:51:00 +08:00
Merge branch 'loc-sens-attn' into loc-sens-attn-new and attention without attention-cum
This commit is contained in:
Eren G committed
commit
9f52833151
6 files changed
+884
-46
No files matched your search
@@ -23,6 +23,7 @@ Checkout [here](https://mycroft.ai/blog/available-voices/#the-human-voice-is-the
|
||||
| [iter-62410](https://drive.google.com/open?id=1pjJNzENL3ZNps9n7k_ktGbpEl6YPIkcZ)| [99d56f7](https://github.com/mozilla/TTS/tree/99d56f7e93ccd7567beb0af8fcbd4d24c48e59e9) | [link](https://soundcloud.com/user-565970875/99d56f7-iter62410 )|First model with plain Tacotron implementation.|
|
||||
| [iter-170K](https://drive.google.com/open?id=16L6JbPXj6MSlNUxEStNn28GiSzi4fu1j) | [e00bc66](https://github.com/mozilla/TTS/tree/e00bc66) |[link](https://soundcloud.com/user-565970875/april-13-2018-07-06pm-e00bc66-iter170k)|More stable and longer trained model.|
|
||||
| Best: [iter-270K](https://drive.google.com/drive/folders/1Q6BKeEkZyxSGsocK2p_mqgzLwlNvbHFJ?usp=sharing)|[256ed63](https://github.com/mozilla/TTS/tree/256ed63)|[link](https://soundcloud.com/user-565970875/sets/samples-1650226)|Stop-Token prediction is added, to detect end of speech.|
|
||||
| Best: [iter-K] | [bla]() | [link]() | Location Sensitive attention |
|
||||
|
||||
## Data
|
||||
Currently TTS provides data loaders for
|
||||
|
||||
+62
-20
@@ -13,8 +13,8 @@ class BahdanauAttention(nn.Module):
|
||||
def forward(self, annots, query):
|
||||
"""
|
||||
Shapes:
|
||||
- query: (batch, 1, dim) or (batch, dim)
|
||||
- annots: (batch, max_time, dim)
|
||||
- query: (batch, 1, dim) or (batch, dim)
|
||||
"""
|
||||
if query.dim() == 2:
|
||||
# insert time-axis for broadcasting
|
||||
@@ -29,31 +29,70 @@ class BahdanauAttention(nn.Module):
|
||||
return alignment.squeeze(-1)
|
||||
|
||||
|
||||
def get_mask_from_lengths(inputs, inputs_lengths):
|
||||
"""Get mask tensor from list of length
|
||||
class LocationSensitiveAttention(nn.Module):
|
||||
"""Location sensitive attention following
|
||||
https://arxiv.org/pdf/1506.07503.pdf"""
|
||||
def __init__(self, annot_dim, query_dim, hidden_dim,
|
||||
kernel_size=7, filters=20):
|
||||
super(LocationSensitiveAttention, self).__init__()
|
||||
self.kernel_size = kernel_size
|
||||
self.filters = filters
|
||||
padding = int((kernel - 1) / 2)
|
||||
self.loc_conv = nn.Conv1d(2, filters,
|
||||
kernel_size=kernel_size, stride=1,
|
||||
padding=padding, bias=False)
|
||||
self.loc_linear = nn.Linear(loc_dim, hidden_dim)
|
||||
self.query_layer = nn.Linear(query_dim, hidden_dim, bias=True)
|
||||
self.annot_layer = nn.Linear(annot_dim, hidden_dim, bias=True)
|
||||
self.v = nn.Linear(hidden_dim, 1, bias=False)
|
||||
|
||||
Args:
|
||||
inputs: Tensor in size (batch, max_time, dim)
|
||||
inputs_lengths: array like
|
||||
"""
|
||||
mask = inputs.data.new(inputs.size(0), inputs.size(1)).byte().zero_()
|
||||
for idx, l in enumerate(inputs_lengths):
|
||||
mask[idx][:l] = 1
|
||||
return ~mask
|
||||
def forward(self, annot, query, loc):
|
||||
"""
|
||||
Shapes:
|
||||
- annot: (batch, max_time, dim)
|
||||
- query: (batch, 1, dim) or (batch, dim)
|
||||
- loc: (batch, 2, max_time)
|
||||
"""
|
||||
if query.dim() == 2:
|
||||
# insert time-axis for broadcasting
|
||||
query = query.unsqueeze(1)
|
||||
loc_conv = self.loc_conv(loc)
|
||||
loc_conv = loc_conv.transpose(1, 2)
|
||||
processed_loc = self.loc_linear(loc_conv)
|
||||
processed_query = self.query_layer(query)
|
||||
processed_annots = self.annot_layer(annot)
|
||||
alignment = self.v(nn.functional.tanh(
|
||||
processed_query + processed_annots + processed_loc))
|
||||
# (batch, max_time)
|
||||
return alignment.squeeze(-1)
|
||||
|
||||
|
||||
class AttentionRNN(nn.Module):
|
||||
def __init__(self, out_dim, annot_dim, memory_dim,
|
||||
score_mask_value=-float("inf")):
|
||||
class AttentionRNNCell(nn.Module):
|
||||
def __init__(self, out_dim, annot_dim, memory_dim, align_model):
|
||||
r"""
|
||||
General Attention RNN wrapper
|
||||
|
||||
Args:
|
||||
out_dim (int): context vector feature dimension.
|
||||
annot_dim (int): annotation vector feature dimension.
|
||||
memory_dim (int): memory vector (decoder autogression) feature dimension.
|
||||
align_model (str): 'b' for Bahdanau, 'ls' Location Sensitive alignment.
|
||||
"""
|
||||
super(AttentionRNN, self).__init__()
|
||||
self.align_model = align_model
|
||||
self.rnn_cell = nn.GRUCell(out_dim + memory_dim, out_dim)
|
||||
self.alignment_model = BahdanauAttention(annot_dim, out_dim, out_dim)
|
||||
self.score_mask_value = score_mask_value
|
||||
# pick bahdanau or location sensitive attention
|
||||
if align_model == 'b':
|
||||
self.alignment_model = BahdanauAttention(annot_dim, out_dim, out_dim)
|
||||
if align_model == 'ls':
|
||||
self.alignment_model = LocationSensitiveAttention(annot_dim, out_dim, out_dim)
|
||||
else:
|
||||
raise RuntimeError(" Wrong alignment model name: {}. Use\
|
||||
'b' (Bahdanau) or 'ls' (Location Sensitive).".format(align_model))
|
||||
|
||||
|
||||
def forward(self, memory, context, rnn_state, annotations,
|
||||
mask=None, annotations_lengths=None):
|
||||
if annotations_lengths is not None and mask is None:
|
||||
mask = get_mask_from_lengths(annotations, annotations_lengths)
|
||||
attention_vec, mask=None, annotations_lengths=None):
|
||||
# Concat input query and previous context context
|
||||
rnn_input = torch.cat((memory, context), -1)
|
||||
# Feed it to RNN
|
||||
@@ -62,7 +101,10 @@ class AttentionRNN(nn.Module):
|
||||
# Alignment
|
||||
# (batch, max_time)
|
||||
# e_{ij} = a(s_{i-1}, h_j)
|
||||
alignment = self.alignment_model(annotations, rnn_output)
|
||||
if self.align_model is 'b':
|
||||
alignment = self.alignment_model(annotations, rnn_output)
|
||||
else:
|
||||
alignment = self.alignment_model(annotations, rnn_output, attention_vec)
|
||||
# TODO: needs recheck.
|
||||
if mask is not None:
|
||||
mask = mask.view(query.size(0), -1)
|
||||
|
||||
+3
-3
@@ -17,10 +17,10 @@ def _sequence_mask(sequence_length, max_len=None):
|
||||
return seq_range_expand < seq_length_expand
|
||||
|
||||
|
||||
class L1LossMasked(nn.Module):
|
||||
class L2LossMasked(nn.Module):
|
||||
|
||||
def __init__(self):
|
||||
super(L1LossMasked, self).__init__()
|
||||
super(L2LossMasked, self).__init__()
|
||||
|
||||
def forward(self, input, target, length):
|
||||
"""
|
||||
@@ -44,7 +44,7 @@ class L1LossMasked(nn.Module):
|
||||
# target_flat: (batch * max_len, dim)
|
||||
target_flat = target.view(-1, target.shape[-1])
|
||||
# losses_flat: (batch * max_len, dim)
|
||||
losses_flat = functional.l1_loss(input, target_flat, size_average=False,
|
||||
losses_flat = functional.mse_loss(input, target_flat, size_average=False,
|
||||
reduce=False)
|
||||
# losses: (batch, max_len, dim)
|
||||
losses = losses_flat.view(*target.size())
|
||||
|
||||
+34
-22
@@ -1,9 +1,7 @@
|
||||
# coding: utf-8
|
||||
import torch
|
||||
from torch import nn
|
||||
from .attention import AttentionRNN
|
||||
from .attention import get_mask_from_lengths
|
||||
|
||||
from .attention import AttentionRNNCell
|
||||
|
||||
class Prenet(nn.Module):
|
||||
r""" Prenet as explained at https://arxiv.org/abs/1703.10135.
|
||||
@@ -12,7 +10,7 @@ class Prenet(nn.Module):
|
||||
Args:
|
||||
in_features (int): size of the input vector
|
||||
out_features (int or list): size of each output sample.
|
||||
If it is a list, for each value, there is created a new layer.
|
||||
If it is a list, for each value, there is created a new layer.
|
||||
"""
|
||||
|
||||
def __init__(self, in_features, out_features=[256, 128]):
|
||||
@@ -162,7 +160,7 @@ class CBHG(nn.Module):
|
||||
x = highway(x)
|
||||
# (B, T_in, in_features*2)
|
||||
# TODO: replace GRU with convolution as in Deep Voice 3
|
||||
self.gru.flatten_parameters()
|
||||
# self.gru.flatten_parameters()
|
||||
outputs, _ = self.gru(x)
|
||||
return outputs
|
||||
|
||||
@@ -195,7 +193,6 @@ class Decoder(nn.Module):
|
||||
in_features (int): input vector (encoder output) sample size.
|
||||
memory_dim (int): memory vector (prev. time-step output) sample size.
|
||||
r (int): number of outputs per time step.
|
||||
eps (float): threshold for detecting the end of a sentence.
|
||||
"""
|
||||
|
||||
def __init__(self, in_features, memory_dim, r):
|
||||
@@ -205,8 +202,8 @@ class Decoder(nn.Module):
|
||||
self.memory_dim = memory_dim
|
||||
# memory -> |Prenet| -> processed_memory
|
||||
self.prenet = Prenet(memory_dim * r, out_features=[256, 128])
|
||||
# processed_inputs, processed_memory -> |Attention| -> Attention, Alignment, RNN_State
|
||||
self.attention_rnn = AttentionRNN(256, in_features, 128)
|
||||
# processed_inputs, processed_memory -> |Attention| -> Attention, attention, RNN_State
|
||||
self.attention_rnn = AttentionRNNCell(256, in_features, 128, align_model='ls')
|
||||
# (processed_memory | attention context) -> |Linear| -> decoder_RNN_input
|
||||
self.project_to_decoder_in = nn.Linear(256+in_features, 256)
|
||||
# decoder_RNN_input -> |RNN| -> RNN_state
|
||||
@@ -234,6 +231,7 @@ class Decoder(nn.Module):
|
||||
- memory: batch x #mel_specs x mel_spec_dim
|
||||
"""
|
||||
B = inputs.size(0)
|
||||
T = inputs.size(1)
|
||||
# Run greedy decoding if memory is None
|
||||
greedy = not self.training
|
||||
if memory is not None:
|
||||
@@ -243,19 +241,22 @@ class Decoder(nn.Module):
|
||||
" !! Dimension mismatch {} vs {} * {}".format(memory.size(-1),
|
||||
self.memory_dim, self.r)
|
||||
T_decoder = memory.size(1)
|
||||
# go frame - 0 frames tarting the sequence
|
||||
# go frame as zeros matrix
|
||||
initial_memory = inputs.data.new(B, self.memory_dim * self.r).zero_()
|
||||
# Init decoder states
|
||||
# decoder states
|
||||
attention_rnn_hidden = inputs.data.new(B, 256).zero_()
|
||||
decoder_rnn_hiddens = [inputs.data.new(B, 256).zero_()
|
||||
for _ in range(len(self.decoder_rnns))]
|
||||
current_context_vec = inputs.data.new(B, 256).zero_()
|
||||
stopnet_rnn_hidden = inputs.data.new(B, self.r * self.memory_dim).zero_()
|
||||
# attention states
|
||||
attention = inputs.data.new(B, T).zero_()
|
||||
# attention_cum = inputs.data.new(B, T).zero_()
|
||||
# Time first (T_decoder, B, memory_dim)
|
||||
if memory is not None:
|
||||
memory = memory.transpose(0, 1)
|
||||
outputs = []
|
||||
alignments = []
|
||||
attentions = []
|
||||
stop_tokens = []
|
||||
t = 0
|
||||
memory_input = initial_memory
|
||||
@@ -268,8 +269,12 @@ class Decoder(nn.Module):
|
||||
# Prenet
|
||||
processed_memory = self.prenet(memory_input)
|
||||
# Attention RNN
|
||||
attention_rnn_hidden, current_context_vec, alignment = self.attention_rnn(
|
||||
processed_memory, current_context_vec, attention_rnn_hidden, inputs)
|
||||
# attention_cat = torch.cat((attention.unsqueeze(1),
|
||||
# attention_cum.unsqueeze(1)),
|
||||
# dim=1)
|
||||
attention_rnn_hidden, current_context_vec, attention = self.attention_rnn(
|
||||
processed_memory, current_context_vec, attention_rnn_hidden, inputs, attention)
|
||||
# attention_cum += attention
|
||||
# Concat RNN output and attention context vector
|
||||
decoder_input = self.project_to_decoder_in(
|
||||
torch.cat((attention_rnn_hidden, current_context_vec), -1))
|
||||
@@ -286,14 +291,14 @@ class Decoder(nn.Module):
|
||||
# predict stop token
|
||||
stop_token, stopnet_rnn_hidden = self.stopnet(stop_input, stopnet_rnn_hidden)
|
||||
outputs += [output]
|
||||
alignments += [alignment]
|
||||
attentions += [attention]
|
||||
stop_tokens += [stop_token]
|
||||
t += 1
|
||||
if (not greedy and self.training) or (greedy and memory is not None):
|
||||
if t >= T_decoder:
|
||||
break
|
||||
else:
|
||||
if t > inputs.shape[1]/2 and stop_token > 0.8:
|
||||
if t > inputs.shape[1]/2 and stop_token > 0.6:
|
||||
break
|
||||
elif t > self.max_decoder_steps:
|
||||
print(" !! Decoder stopped with 'max_decoder_steps'. \
|
||||
@@ -301,28 +306,35 @@ class Decoder(nn.Module):
|
||||
break
|
||||
assert greedy or len(outputs) == T_decoder
|
||||
# Back to batch first
|
||||
alignments = torch.stack(alignments).transpose(0, 1)
|
||||
attentions = torch.stack(attentions).transpose(0, 1)
|
||||
outputs = torch.stack(outputs).transpose(0, 1).contiguous()
|
||||
stop_tokens = torch.stack(stop_tokens).transpose(0, 1)
|
||||
return outputs, alignments, stop_tokens
|
||||
return outputs, attentions, stop_tokens
|
||||
|
||||
|
||||
|
||||
class StopNet(nn.Module):
|
||||
r"""
|
||||
Predicting stop-token in decoder.
|
||||
|
||||
|
||||
Args:
|
||||
r (int): number of output frames of the network.
|
||||
memory_dim (int): feature dimension for each output frame.
|
||||
"""
|
||||
|
||||
|
||||
def __init__(self, r, memory_dim):
|
||||
r"""
|
||||
Predicts the stop token to stop the decoder at testing time
|
||||
|
||||
Args:
|
||||
r (int): number of network output frames.
|
||||
memory_dim (int): single feature dim of a single network output frame.
|
||||
"""
|
||||
super(StopNet, self).__init__()
|
||||
self.rnn = nn.GRUCell(memory_dim * r, memory_dim * r)
|
||||
self.relu = nn.ReLU()
|
||||
self.linear = nn.Linear(r * memory_dim, 1)
|
||||
self.sigmoid = nn.Sigmoid()
|
||||
|
||||
|
||||
def forward(self, inputs, rnn_hidden):
|
||||
"""
|
||||
Args:
|
||||
@@ -333,4 +345,4 @@ class StopNet(nn.Module):
|
||||
outputs = self.relu(rnn_hidden)
|
||||
outputs = self.linear(outputs)
|
||||
outputs = self.sigmoid(outputs)
|
||||
return outputs, rnn_hidden
|
||||
return outputs, rnn_hidden
|
||||
@@ -0,0 +1,783 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 1,
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Populating the interactive namespace from numpy and matplotlib\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"name": "stderr",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"/home/erogol/miniconda3/envs/pytorch4/lib/python3.6/site-packages/IPython/core/magics/pylab.py:160: UserWarning: pylab import has clobbered these variables: ['plt']\n",
|
||||
"`%matplotlib` prevents importing * from pylab and numpy\n",
|
||||
" \"\\n`%matplotlib` prevents importing * from pylab and numpy\"\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"%load_ext autoreload\n",
|
||||
"%autoreload 2\n",
|
||||
"import os\n",
|
||||
"import sys\n",
|
||||
"import io\n",
|
||||
"import torch \n",
|
||||
"import time\n",
|
||||
"import numpy as np\n",
|
||||
"from collections import OrderedDict\n",
|
||||
"from matplotlib import pylab as plt\n",
|
||||
"\n",
|
||||
"%pylab inline\n",
|
||||
"rcParams[\"figure.figsize\"] = (16,5)\n",
|
||||
"sys.path.append('/home/erogol/projects/')\n",
|
||||
"\n",
|
||||
"import librosa\n",
|
||||
"import librosa.display\n",
|
||||
"\n",
|
||||
"from TTS.models.tacotron import Tacotron \n",
|
||||
"from TTS.layers import *\n",
|
||||
"from TTS.utils.data import *\n",
|
||||
"from TTS.utils.audio import AudioProcessor\n",
|
||||
"from TTS.utils.generic_utils import load_config\n",
|
||||
"from TTS.utils.text import text_to_sequence\n",
|
||||
"\n",
|
||||
"import IPython\n",
|
||||
"from IPython.display import Audio\n",
|
||||
"from utils import *"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 2,
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"\u001b[0m\u001b[01;31mcheckpoint_33464.pth.tar\u001b[0m \u001b[01;31mcheckpoint_36096.pth.tar\u001b[0m\r\n",
|
||||
"\u001b[01;31mcheckpoint_33840.pth.tar\u001b[0m \u001b[01;31mcheckpoint_36472.pth.tar\u001b[0m\r\n",
|
||||
"\u001b[01;31mcheckpoint_34216.pth.tar\u001b[0m \u001b[01;31mcheckpoint_36848.pth.tar\u001b[0m\r\n",
|
||||
"\u001b[01;31mcheckpoint_34592.pth.tar\u001b[0m \u001b[01;31mcheckpoint_37224.pth.tar\u001b[0m\r\n",
|
||||
"\u001b[01;31mcheckpoint_34968.pth.tar\u001b[0m \u001b[01;34mcheckpoints\u001b[0m/\r\n",
|
||||
"\u001b[01;31mcheckpoint_35344.pth.tar\u001b[0m config.json\r\n",
|
||||
"\u001b[01;31mcheckpoint_35720.pth.tar\u001b[0m events.out.tfevents.1526635608.mlc1\r\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"ls /data/shared/erogol_models/May-18-2018_02:26AM-los_sen_attn-debug "
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 3,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"def tts(model, text, CONFIG, use_cuda, ap, figures=True):\n",
|
||||
" t_1 = time.time()\n",
|
||||
" waveform, alignment, spectrogram, stop_tokens = create_speech(model, text, CONFIG, use_cuda, ap) \n",
|
||||
" print(\" > Run-time: {}\".format(time.time() - t_1))\n",
|
||||
" if figures: \n",
|
||||
" visualize(alignment, spectrogram, stop_tokens, CONFIG) \n",
|
||||
" IPython.display.display(Audio(waveform, rate=CONFIG.sample_rate)) \n",
|
||||
" return alignment, spectrogram, stop_tokens"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 5,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# Set constants\n",
|
||||
"ROOT_PATH = '/data/shared/erogol_models/May-18-2018_02:26AM-los_sen_attn-debug/'\n",
|
||||
"MODEL_PATH = ROOT_PATH + '/checkpoint_37224.pth.tar'\n",
|
||||
"CONFIG_PATH = ROOT_PATH + '/config.json'\n",
|
||||
"OUT_FOLDER = ROOT_PATH + '/test/'\n",
|
||||
"CONFIG = load_config(CONFIG_PATH)\n",
|
||||
"use_cuda = True"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 6,
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
" | > Number of characters : 149\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"data": {
|
||||
"text/plain": [
|
||||
"Tacotron(\n",
|
||||
" (embedding): Embedding(149, 256)\n",
|
||||
" (encoder): Encoder(\n",
|
||||
" (prenet): Prenet(\n",
|
||||
" (layers): ModuleList(\n",
|
||||
" (0): Linear(in_features=256, out_features=256, bias=True)\n",
|
||||
" (1): Linear(in_features=256, out_features=128, bias=True)\n",
|
||||
" )\n",
|
||||
" (relu): ReLU()\n",
|
||||
" (dropout): Dropout(p=0.5)\n",
|
||||
" )\n",
|
||||
" (cbhg): CBHG(\n",
|
||||
" (relu): ReLU()\n",
|
||||
" (conv1d_banks): ModuleList(\n",
|
||||
" (0): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(1,), stride=(1,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (1): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(2,), stride=(1,), padding=(1,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (2): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(3,), stride=(1,), padding=(1,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (3): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(4,), stride=(1,), padding=(2,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (4): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(5,), stride=(1,), padding=(2,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (5): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(6,), stride=(1,), padding=(3,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (6): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(7,), stride=(1,), padding=(3,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (7): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(8,), stride=(1,), padding=(4,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (8): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(9,), stride=(1,), padding=(4,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (9): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(10,), stride=(1,), padding=(5,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (10): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(11,), stride=(1,), padding=(5,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (11): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(12,), stride=(1,), padding=(6,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (12): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(13,), stride=(1,), padding=(6,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (13): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(14,), stride=(1,), padding=(7,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (14): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(15,), stride=(1,), padding=(7,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (15): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(16,), stride=(1,), padding=(8,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" )\n",
|
||||
" (max_pool1d): MaxPool1d(kernel_size=2, stride=1, padding=1, dilation=1, ceil_mode=False)\n",
|
||||
" (conv1d_projections): ModuleList(\n",
|
||||
" (0): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(2048, 128, kernel_size=(3,), stride=(1,), padding=(1,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (1): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(128, 128, kernel_size=(3,), stride=(1,), padding=(1,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(128, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" )\n",
|
||||
" )\n",
|
||||
" (pre_highway): Linear(in_features=128, out_features=128, bias=False)\n",
|
||||
" (highways): ModuleList(\n",
|
||||
" (0): Highway(\n",
|
||||
" (H): Linear(in_features=128, out_features=128, bias=True)\n",
|
||||
" (T): Linear(in_features=128, out_features=128, bias=True)\n",
|
||||
" (relu): ReLU()\n",
|
||||
" (sigmoid): Sigmoid()\n",
|
||||
" )\n",
|
||||
" (1): Highway(\n",
|
||||
" (H): Linear(in_features=128, out_features=128, bias=True)\n",
|
||||
" (T): Linear(in_features=128, out_features=128, bias=True)\n",
|
||||
" (relu): ReLU()\n",
|
||||
" (sigmoid): Sigmoid()\n",
|
||||
" )\n",
|
||||
" (2): Highway(\n",
|
||||
" (H): Linear(in_features=128, out_features=128, bias=True)\n",
|
||||
" (T): Linear(in_features=128, out_features=128, bias=True)\n",
|
||||
" (relu): ReLU()\n",
|
||||
" (sigmoid): Sigmoid()\n",
|
||||
" )\n",
|
||||
" (3): Highway(\n",
|
||||
" (H): Linear(in_features=128, out_features=128, bias=True)\n",
|
||||
" (T): Linear(in_features=128, out_features=128, bias=True)\n",
|
||||
" (relu): ReLU()\n",
|
||||
" (sigmoid): Sigmoid()\n",
|
||||
" )\n",
|
||||
" )\n",
|
||||
" (gru): GRU(128, 128, batch_first=True, bidirectional=True)\n",
|
||||
" )\n",
|
||||
" )\n",
|
||||
" (decoder): Decoder(\n",
|
||||
" (prenet): Prenet(\n",
|
||||
" (layers): ModuleList(\n",
|
||||
" (0): Linear(in_features=400, out_features=256, bias=True)\n",
|
||||
" (1): Linear(in_features=256, out_features=128, bias=True)\n",
|
||||
" )\n",
|
||||
" (relu): ReLU()\n",
|
||||
" (dropout): Dropout(p=0.5)\n",
|
||||
" )\n",
|
||||
" (attention_rnn): AttentionRNN(\n",
|
||||
" (rnn_cell): GRUCell(384, 256)\n",
|
||||
" (alignment_model): LocationSensitiveAttention(\n",
|
||||
" (loc_conv): Conv1d(2, 32, kernel_size=(31,), stride=(1,), padding=(15,), bias=False)\n",
|
||||
" (loc_linear): Linear(in_features=32, out_features=256, bias=True)\n",
|
||||
" (query_layer): Linear(in_features=256, out_features=256, bias=True)\n",
|
||||
" (annot_layer): Linear(in_features=256, out_features=256, bias=True)\n",
|
||||
" (v): Linear(in_features=256, out_features=1, bias=False)\n",
|
||||
" )\n",
|
||||
" )\n",
|
||||
" (project_to_decoder_in): Linear(in_features=512, out_features=256, bias=True)\n",
|
||||
" (decoder_rnns): ModuleList(\n",
|
||||
" (0): GRUCell(256, 256)\n",
|
||||
" (1): GRUCell(256, 256)\n",
|
||||
" )\n",
|
||||
" (proj_to_mel): Linear(in_features=256, out_features=400, bias=True)\n",
|
||||
" (stopnet): StopNet(\n",
|
||||
" (rnn): GRUCell(400, 400)\n",
|
||||
" (relu): ReLU()\n",
|
||||
" (linear): Linear(in_features=400, out_features=1, bias=True)\n",
|
||||
" (sigmoid): Sigmoid()\n",
|
||||
" )\n",
|
||||
" )\n",
|
||||
" (postnet): CBHG(\n",
|
||||
" (relu): ReLU()\n",
|
||||
" (conv1d_banks): ModuleList(\n",
|
||||
" (0): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(80, 80, kernel_size=(1,), stride=(1,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(80, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (1): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(80, 80, kernel_size=(2,), stride=(1,), padding=(1,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(80, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (2): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(80, 80, kernel_size=(3,), stride=(1,), padding=(1,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(80, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (3): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(80, 80, kernel_size=(4,), stride=(1,), padding=(2,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(80, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (4): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(80, 80, kernel_size=(5,), stride=(1,), padding=(2,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(80, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (5): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(80, 80, kernel_size=(6,), stride=(1,), padding=(3,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(80, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (6): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(80, 80, kernel_size=(7,), stride=(1,), padding=(3,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(80, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (7): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(80, 80, kernel_size=(8,), stride=(1,), padding=(4,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(80, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" )\n",
|
||||
" (max_pool1d): MaxPool1d(kernel_size=2, stride=1, padding=1, dilation=1, ceil_mode=False)\n",
|
||||
" (conv1d_projections): ModuleList(\n",
|
||||
" (0): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(640, 256, kernel_size=(3,), stride=(1,), padding=(1,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(256, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" (activation): ReLU()\n",
|
||||
" )\n",
|
||||
" (1): BatchNormConv1d(\n",
|
||||
" (conv1d): Conv1d(256, 80, kernel_size=(3,), stride=(1,), padding=(1,), bias=False)\n",
|
||||
" (bn): BatchNorm1d(80, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n",
|
||||
" )\n",
|
||||
" )\n",
|
||||
" (pre_highway): Linear(in_features=80, out_features=80, bias=False)\n",
|
||||
" (highways): ModuleList(\n",
|
||||
" (0): Highway(\n",
|
||||
" (H): Linear(in_features=80, out_features=80, bias=True)\n",
|
||||
" (T): Linear(in_features=80, out_features=80, bias=True)\n",
|
||||
" (relu): ReLU()\n",
|
||||
" (sigmoid): Sigmoid()\n",
|
||||
" )\n",
|
||||
" (1): Highway(\n",
|
||||
" (H): Linear(in_features=80, out_features=80, bias=True)\n",
|
||||
" (T): Linear(in_features=80, out_features=80, bias=True)\n",
|
||||
" (relu): ReLU()\n",
|
||||
" (sigmoid): Sigmoid()\n",
|
||||
" )\n",
|
||||
" (2): Highway(\n",
|
||||
" (H): Linear(in_features=80, out_features=80, bias=True)\n",
|
||||
" (T): Linear(in_features=80, out_features=80, bias=True)\n",
|
||||
" (relu): ReLU()\n",
|
||||
" (sigmoid): Sigmoid()\n",
|
||||
" )\n",
|
||||
" (3): Highway(\n",
|
||||
" (H): Linear(in_features=80, out_features=80, bias=True)\n",
|
||||
" (T): Linear(in_features=80, out_features=80, bias=True)\n",
|
||||
" (relu): ReLU()\n",
|
||||
" (sigmoid): Sigmoid()\n",
|
||||
" )\n",
|
||||
" )\n",
|
||||
" (gru): GRU(80, 80, batch_first=True, bidirectional=True)\n",
|
||||
" )\n",
|
||||
" (last_linear): Linear(in_features=160, out_features=1025, bias=True)\n",
|
||||
")"
|
||||
]
|
||||
},
|
||||
"execution_count": 6,
|
||||
"metadata": {},
|
||||
"output_type": "execute_result"
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"# load the model\n",
|
||||
"model = Tacotron(CONFIG.embedding_size, CONFIG.num_freq, CONFIG.num_mels, CONFIG.r)\n",
|
||||
"\n",
|
||||
"# load the audio processor\n",
|
||||
"\n",
|
||||
"ap = AudioProcessor(CONFIG.sample_rate, CONFIG.num_mels, CONFIG.min_level_db,\n",
|
||||
" CONFIG.frame_shift_ms, CONFIG.frame_length_ms, CONFIG.preemphasis,\n",
|
||||
" CONFIG.ref_level_db, CONFIG.num_freq, CONFIG.power, griffin_lim_iters=30) \n",
|
||||
"\n",
|
||||
"\n",
|
||||
"# load model state\n",
|
||||
"if use_cuda:\n",
|
||||
" cp = torch.load(MODEL_PATH)\n",
|
||||
"else:\n",
|
||||
" cp = torch.load(MODEL_PATH, map_location=lambda storage, loc: storage)\n",
|
||||
"\n",
|
||||
"# load the model\n",
|
||||
"model.load_state_dict(cp['model'])\n",
|
||||
"if use_cuda:\n",
|
||||
" model.cuda()\n",
|
||||
"model.eval()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### EXAMPLES FROM TRAINING SET"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 8,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import pandas as pd\n",
|
||||
"df = pd.read_csv('/data/shared/KeithIto/LJSpeech-1.0/metadata_val.csv', delimiter='|')"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 9,
|
||||
"metadata": {
|
||||
"scrolled": false
|
||||
},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Hosty took the necessary steps to have the Dallas office of the FBI, rather than the New Orleans office, reestablished as the office with principal responsibility.\n",
|
||||
" > Run-time: 9.547307014465332\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"data": {
|
||||
"text/html": [
|
||||
"\n",
|
||||
" <audio controls=\"controls\" >\n",
|
||||
" <source src=\"data:audio/wav;base64,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 truncated
|
||||
" Your browser does not support the audio element.\n",
|
||||
" </audio>\n",
|
||||
" "
|
||||
],
|
||||
"text/plain": [
|
||||
"<IPython.lib.display.Audio object>"
|
||||
]
|
||||
},
|
||||
"metadata": {},
|
||||
"output_type": "display_data"
|
||||
},
|
||||
{
|
||||
"data": {
|
||||
"image/png": "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 truncated
|
||||
"text/plain": [
|
||||
"<Figure size 1152x1728 with 5 Axes>"
|
||||
]
|
||||
},
|
||||
"metadata": {},
|
||||
"output_type": "display_data"
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"sentence = df.iloc[175, 1]\n",
|
||||
"print(sentence)\n",
|
||||
"model.decoder.max_decoder_steps = 250\n",
|
||||
"align, spec, stop_tokens = tts(model, sentence, CONFIG, use_cuda, ap)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### Comparision with https://mycroft.ai/blog/available-voices/"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 10,
|
||||
"metadata": {
|
||||
"scrolled": false
|
||||
},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
" > Run-time: 5.965982913970947\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"data": {
|
||||
"text/html": [
|
||||
"\n",
|
||||
" <audio controls=\"controls\" >\n",
|
||||
" <source src=\"data:audio/wav;base64,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 truncated
|
||||
" Your browser does not support the audio element.\n",
|
||||
" </audio>\n",
|
||||
" "
|
||||
],
|
||||
"text/plain": [
|
||||
"<IPython.lib.display.Audio object>"
|
||||
]
|
||||
},
|
||||
"metadata": {},
|
||||
"output_type": "display_data"
|
||||
},
|
||||
{
|
||||
"data": {
|
||||
"image/png": "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 truncated
|
||||
"text/plain": [
|
||||
"<Figure size 1152x1728 with 5 Axes>"
|
||||
]
|
||||
},
|
||||
"metadata": {},
|
||||
"output_type": "display_data"
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"sentence = \"It took me quite a long time to develop a voice, and now that I have it I'm not going to be silent.\"\n",
|
||||
"model.decoder.max_decoder_steps = 250\n",
|
||||
"align, spec, stop_tokens = tts(model, sentence, CONFIG, use_cuda, ap, figures=True)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 11,
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
" > Run-time: 1.3593213558197021\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"data": {
|
||||
"text/html": [
|
||||
"\n",
|
||||
" <audio controls=\"controls\" >\n",
|
||||
" <source src=\"data:audio/wav;base64,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 truncated
|
||||
" Your browser does not support the audio element.\n",
|
||||
" </audio>\n",
|
||||
" "
|
||||
],
|
||||
"text/plain": [
|
||||
"<IPython.lib.display.Audio object>"
|
||||
]
|
||||
},
|
||||
"metadata": {},
|
||||
"output_type": "display_data"
|
||||
},
|
||||
{
|
||||
"data": {
|
||||
"image/png": "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 truncated
|
||||
"text/plain": [
|
||||
"<Figure size 1152x1728 with 5 Axes>"
|
||||
]
|
||||
},
|
||||
"metadata": {},
|
||||
"output_type": "display_data"
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"sentence = \"Be a voice,not an echo.\" # 'echo' is not in training set. \n",
|
||||
"align, spec, stop_tokens = tts(model, sentence, CONFIG, use_cuda, ap)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 12,
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
" > Run-time: 3.3218443393707275\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"data": {
|
||||
"text/html": [
|
||||
"\n",
|
||||
" <audio controls=\"controls\" >\n",
|
||||
" <source src=\"data:audio/wav;base64,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 truncated
|
||||
" Your browser does not support the audio element.\n",
|
||||
" </audio>\n",
|
||||
" "
|
||||
],
|
||||
"text/plain": [
|
||||
"<IPython.lib.display.Audio object>"
|
||||
]
|
||||
},
|
||||
"metadata": {},
|
||||
"output_type": "display_data"
|
||||
},
|
||||
{
|
||||
"data": {
|
||||
"image/png": "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 truncated
|
||||
"text/plain": [
|
||||
"<Figure size 1152x1728 with 5 Axes>"
|
||||
]
|
||||
},
|
||||
"metadata": {},
|
||||
"output_type": "display_data"
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"sentence = \"The human voice is the most perfect instrument of all.\"\n",
|
||||
"align, spec, stop_tokens = tts(model, sentence, CONFIG, use_cuda, ap)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 13,
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
" > Run-time: 3.3077054023742676\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"data": {
|
||||
"text/html": [
|
||||
"\n",
|
||||
" <audio controls=\"controls\" >\n",
|
||||
" <source src=\"data:audio/wav;base64,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 truncated
|
||||
" Your browser does not support the audio element.\n",
|
||||
" </audio>\n",
|
||||
" "
|
||||
],
|
||||
"text/plain": [
|
||||
"<IPython.lib.display.Audio object>"
|
||||
]
|
||||
},
|
||||
"metadata": {},
|
||||
"output_type": "display_data"
|
||||
},
|
||||
{
|
||||
"data": {
|
||||
"image/png": "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 truncated
|
||||
"text/plain": [
|
||||
"<Figure size 1152x1728 with 5 Axes>"
|
||||
]
|
||||
},
|
||||
"metadata": {},
|
||||
"output_type": "display_data"
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"sentence = \"I'm sorry Dave. I'm afraid I can't do that.\"\n",
|
||||
"align, spec, stop_tokens = tts(model, sentence, CONFIG, use_cuda, ap)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
" > Run-time: 3.859889507293701\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"data": {
|
||||
"text/html": [
|
||||
"\n",
|
||||
" <audio controls=\"controls\" >\n",
|
||||
" <source src=\"data:audio/wav;base64,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 truncated
|
||||
" Your browser does not support the audio element.\n",
|
||||
" </audio>\n",
|
||||
" "
|
||||
],
|
||||
"text/plain": [
|
||||
"<IPython.lib.display.Audio object>"
|
||||
]
|
||||
},
|
||||
"metadata": {},
|
||||
"output_type": "display_data"
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"sentence = \"This cake is great. It's so delicious and moist.\"\n",
|
||||
"align, spec, stop_tokens = tts(model, sentence, CONFIG, use_cuda, ap)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### Comparison with https://keithito.github.io/audio-samples/"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"sentence = \"Generative adversarial network or variational auto-encoder.\"\n",
|
||||
"align, spec, stop_tokens = tts(model, sentence, CONFIG, use_cuda, ap)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"sentence = \"Scientists at the CERN laboratory say they have discovered a new particle.\"\n",
|
||||
"align, spec, stop_tokens = tts(model, sentence, CONFIG, use_cuda, ap)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"sentence = \"here’s a way to measure the acute emotional intelligence that has never gone out of style.\"\n",
|
||||
"align, spec, stop_tokens = tts(model, sentence, CONFIG, use_cuda, ap)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"sentence = \"President Trump met with other leaders at the Group of 20 conference.\"\n",
|
||||
"align, spec, stop_tokens = tts(model, sentence, CONFIG, use_cuda, ap)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"sentence = \"The buses aren't the problem, they actually provide a solution.\"\n",
|
||||
"align, spec, stop_tokens = tts(model, sentence, CONFIG, use_cuda, ap)"
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3",
|
||||
"language": "python",
|
||||
"name": "python3"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3",
|
||||
"version": "3.6.5"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 2
|
||||
}
|
||||
@@ -26,7 +26,7 @@ from utils.model import get_param_size
|
||||
from utils.visual import plot_alignment, plot_spectrogram
|
||||
from datasets.LJSpeech import LJSpeechDataset
|
||||
from models.tacotron import Tacotron
|
||||
from layers.losses import L1LossMasked
|
||||
from layers.losses import L2LossMasked
|
||||
|
||||
torch.manual_seed(1)
|
||||
use_cuda = torch.cuda.is_available()
|
||||
|
||||
Reference in new issue
Block a user