PR for Version Up in upstream repos

README.md

@@ -10,6 +10,12 @@ PyTorch implementation of convolutional networks-based text-to-speech synthesis
 
 Audio samples are available at https://r9y9.github.io/deepvoice3_pytorch/.
 
+## Online TTS demo
+
+A notebook supposed to be executed on https://colab.research.google.com is available:
+
+- [DeepVoice3: Multi-speaker text-to-speech demo](https://colab.research.google.com/github/r9y9/Colaboratory/blob/master/DeepVoice3_multi_speaker_TTS_en_demo.ipynb)
+
 ## Highlights
 
 - Convolutional sequence-to-sequence model with attention for text-to-speech synthesis
@@ -53,9 +59,7 @@ See "Synthesize from a checkpoint" section in the README for how to generate spe
 
 - Python 3
 - CUDA >= 8.0
-- PyTorch >= v0.3
 - TensorFlow >= v1.3
-- [tensorboard-pytorch](https://github.com/lanpa/tensorboard-pytorch) (master)
 - [nnmnkwii](https://github.com/r9y9/nnmnkwii) >= v0.0.11
 - [MeCab](http://taku910.github.io/mecab/) (Japanese only)
 

deepvoice3_pytorch/frontend/text/numbers.py

@@ -1,3 +1,5 @@
+# -*- coding: utf-8 -*-
+
 import inflect
 import re
 

hparams.py

@@ -21,7 +21,7 @@
     # Convenient model builder
     # [deepvoice3, deepvoice3_multispeaker, nyanko]
     # Definitions can be found at deepvoice3_pytorch/builder.py
-    # deepvoice3: DeepVoice3　https://arxiv.org/abs/1710.07654
+    # deepvoice3: DeepVoice3 https://arxiv.org/abs/1710.07654
     # deepvoice3_multispeaker: Multi-speaker version of DeepVoice3
     # nyanko: https://arxiv.org/abs/1710.08969
     builder="deepvoice3",

presets/deepvoice3_vctk.json

@@ -2,7 +2,7 @@
   "name": "deepvoice3",
   "frontend": "en",
   "replace_pronunciation_prob": 0.5,
-  "builder": "deepvoice3",
+  "builder": "deepvoice3_multispeaker",
   "n_speakers": 108,
   "speaker_embed_dim": 16,
   "num_mels": 80,
@@ -62,4 +62,4 @@
   "window_ahead": 3,
   "window_backward": 1,
   "power": 1.4
-}
+}

setup.py

@@ -7,7 +7,7 @@
 import subprocess
 from os.path import exists
 
-version = '0.0.4'
+version = '0.0.5'
 
 # Adapted from https://github.com/pytorch/pytorch
 cwd = os.path.dirname(os.path.abspath(__file__))
@@ -79,6 +79,7 @@ def create_readme_rst():
       install_requires=[
           "numpy",
           "scipy",
+          "torch >= 0.3.0",
           "unidecode",
           "inflect",
           "librosa",

synthesis.py

@@ -78,6 +78,15 @@ def tts(model, text, p=0, speaker_id=None, fast=False):
     return waveform, alignment, spectrogram, mel
 
 
+def _load(checkpoint_path):
+    if use_cuda:
+        checkpoint = torch.load(checkpoint_path)
+    else:
+        checkpoint = torch.load(checkpoint_path,
+                                map_location=lambda storage, loc: storage)
+    return checkpoint
+
+
 if __name__ == "__main__":
     args = docopt(__doc__)
     print("Command line args:\n", args)
@@ -113,13 +122,13 @@ def tts(model, text, p=0, speaker_id=None, fast=False):
 
     # Load checkpoints separately
     if checkpoint_postnet_path is not None and checkpoint_seq2seq_path is not None:
-        checkpoint = torch.load(checkpoint_seq2seq_path)
+        checkpoint = _load(checkpoint_seq2seq_path)
         model.seq2seq.load_state_dict(checkpoint["state_dict"])
-        checkpoint = torch.load(checkpoint_postnet_path)
+        checkpoint = _load(checkpoint_postnet_path)
         model.postnet.load_state_dict(checkpoint["state_dict"])
         checkpoint_name = splitext(basename(checkpoint_seq2seq_path))[0]
     else:
-        checkpoint = torch.load(checkpoint_path)
+        checkpoint = _load(checkpoint_path)
         model.load_state_dict(checkpoint["state_dict"])
         checkpoint_name = splitext(basename(checkpoint_path))[0]
 

train.py

@@ -815,12 +815,21 @@ def build_model():
     return model
 
 
+def _load(checkpoint_path):
+    if use_cuda:
+        checkpoint = torch.load(checkpoint_path)
+    else:
+        checkpoint = torch.load(checkpoint_path,
+                                map_location=lambda storage, loc: storage)
+    return checkpoint
+
+
 def load_checkpoint(path, model, optimizer, reset_optimizer):
     global global_step
     global global_epoch
 
     print("Load checkpoint from: {}".format(path))
-    checkpoint = torch.load(path)
+    checkpoint = _load(path)
     model.load_state_dict(checkpoint["state_dict"])
     if not reset_optimizer:
         optimizer_state = checkpoint["optimizer"]
@@ -834,19 +843,33 @@ def load_checkpoint(path, model, optimizer, reset_optimizer):
 
 
 def _load_embedding(path, model):
-    state = torch.load(path)["state_dict"]
+    state = _load(path)["state_dict"]
     key = "seq2seq.encoder.embed_tokens.weight"
     model.seq2seq.encoder.embed_tokens.weight.data = state[key]
 
-
 # https://discuss.pytorch.org/t/how-to-load-part-of-pre-trained-model/1113/3
+
+
 def restore_parts(path, model):
     print("Restore part of the model from: {}".format(path))
-    state = torch.load(path)["state_dict"]
+    state = _load(path)["state_dict"]
     model_dict = model.state_dict()
     valid_state_dict = {k: v for k, v in state.items() if k in model_dict}
-    model_dict.update(valid_state_dict)
-    model.load_state_dict(model_dict)
+
+    try:
+        model_dict.update(valid_state_dict)
+        model.load_state_dict(model_dict)
+    except RuntimeError as e:
+        # there should be invalid size of weight(s), so load them per parameter
+        print(str(e))
+        model_dict = model.state_dict()
+        for k, v in valid_state_dict.items():
+            model_dict[k] = v
+            try:
+                model.load_state_dict(model_dict)
+            except RuntimeError as e:
+                print(str(e))
+                warn("{}: may contain invalid size of weight. skipping...".format(k))
 
 
 if __name__ == "__main__":
@@ -951,7 +974,8 @@ def restore_parts(path, model):
     # Setup summary writer for tensorboard
     if log_event_path is None:
         if platform.system() == "Windows":
-            log_event_path = "log/run-test" + str(datetime.now()).replace(" ", "_").replace(":","_")
+            log_event_path = "log/run-test" + \
+                str(datetime.now()).replace(" ", "_").replace(":", "_")
         else:
             log_event_path = "log/run-test" + str(datetime.now()).replace(" ", "_")
     print("Los event path: {}".format(log_event_path))