Merge branch 'add_TranTTS' into 'master'

modified load checkpoint in train

See merge request !2

parakeet/models/fastspeech/config/fastapeech.yaml

@@ -1,41 +0,0 @@
-audio:
-  num_mels: 80
-  n_fft: 2048
-  sr: 22050
-  preemphasis: 0.97
-  hop_length: 275
-  win_length: 1102
-  power: 1.2
-  min_level_db: -100
-  ref_level_db: 20
-  outputs_per_step: 1
-
-encoder_n_layer: 6
-encoder_head: 2
-encoder_conv1d_filter_size: 1536
-max_sep_len: 2048
-encoder_output_size: 384
-word_vec_dim: 384
-decoder_n_layer: 6
-decoder_head: 2
-decoder_conv1d_filter_size: 1536
-decoder_output_size: 384
-d_model: 384
-duration_predictor_output_size: 256
-duration_predictor_filter_size: 3
-fft_conv1d_filter: 3
-fft_conv1d_padding: 1
-
-
-batch_size: 32
-epochs: 10000
-lr: 0.001
-save_step: 500
-image_step: 2000
-use_gpu: False
-use_data_parallel: False
-
-data_path: ../../../dataset/LJSpeech-1.1
-transtts_path: ./checkpoint
-transformer_step: 70000
-log_dir: ./log

parakeet/models/transformerTTS/config/train_postnet.yaml

@@ -25,3 +25,5 @@ use_data_parallel: True
 data_path: ../../../dataset/LJSpeech-1.1
 save_path: ./checkpoint
 log_dir: ./log
+#checkpoint_path: ./checkpoint
+#transformer_step: 27000

parakeet/models/transformerTTS/config/train_transformer.yaml

@@ -28,6 +28,7 @@ use_data_parallel: False
 data_path: ../../../dataset/LJSpeech-1.1
 save_path: ./checkpoint
 log_dir: ./log
-#checkpoint_path: ./checkpoint/transformer/1
+#checkpoint_path: ./checkpoint
+#transformer_step: 70000
 
    

parakeet/models/transformerTTS/train_postnet.py

@@ -25,6 +25,9 @@ class MyDataParallel(dg.parallel.DataParallel):
             return getattr(
                 object.__getattribute__(self, "_sub_layers")["_layers"], key)
 
+def load_checkpoint(step, model_path):
+    model_dict, opti_dict = fluid.dygraph.load_dygraph(os.path.join(model_path, step))
+    return model_dict, opti_dict
 
 def main(cfg):
     
@@ -55,9 +58,10 @@ def main(cfg):
 
 
         if cfg.checkpoint_path is not None:
-            model_dict, opti_dict = fluid.dygraph.load_dygraph(cfg.checkpoint_path)
+            model_dict, opti_dict = load_checkpoint(str(cfg.postnet_step), os.path.join(cfg.checkpoint_path, "postnet"))
             model.set_dict(model_dict)
             optimizer.set_dict(opti_dict)
+            global_step = cfg.postnet_step
             print("load checkpoint!!!")
 
         if cfg.use_data_parallel:

parakeet/models/transformerTTS/train_transformer.py

@@ -29,6 +29,10 @@ class MyDataParallel(dg.parallel.DataParallel):
             return getattr(
                 object.__getattribute__(self, "_sub_layers")["_layers"], key)
 
+def load_checkpoint(step, model_path):
+    model_dict, opti_dict = fluid.dygraph.load_dygraph(os.path.join(model_path, step))
+    return model_dict, opti_dict
+
 
 def main(cfg):
     local_rank = dg.parallel.Env().local_rank if cfg.use_data_parallel else 0
@@ -62,9 +66,10 @@ def main(cfg):
         reader = LJSpeechLoader(cfg, nranks, local_rank).reader()
         
         if cfg.checkpoint_path is not None:
-            model_dict, opti_dict = fluid.dygraph.load_dygraph(cfg.checkpoint_path)
+            model_dict, opti_dict = load_checkpoint(str(cfg.transformer_step), os.path.join(cfg.checkpoint_path, "transformer"))
             model.set_dict(model_dict)
             optimizer.set_dict(opti_dict)
+            global_step = cfg.transformer_step
             print("load checkpoint!!!")
 
         if cfg.use_data_parallel: