move data.py and rename some files

.gitignore

@@ -132,7 +132,7 @@ dmypy.json
 .pyre/
 
 # Shell, vim, and output folder
-*.sh
+#*.sh
 *.swp
 runs
 syn_audios

examples/fastspeech/train.py

@@ -13,10 +13,12 @@ from tensorboardX import SummaryWriter
 import paddle.fluid.dygraph as dg
 import paddle.fluid.layers as layers
 import paddle.fluid as fluid
-from parakeet.models.dataloader.ljspeech import LJSpeechLoader
-from parakeet.models.transformer_tts.transformerTTS import TransformerTTS
+from parakeet.models.transformer_tts.transformer_tts import TransformerTTS
 from parakeet.models.fastspeech.fastspeech import FastSpeech
 from parakeet.models.fastspeech.utils import get_alignment
+import sys
+sys.path.append("../transformer_tts")
+from data import LJSpeechLoader
 
 def load_checkpoint(step, model_path):
     model_dict, opti_dict = fluid.dygraph.load_dygraph(os.path.join(model_path, step))

examples/fastspeech/train.sh

-
 # train model
 # if you wish to resume from an exists model, uncomment --checkpoint_path and --fastspeech_step
 #CUDA_VISIBLE_DEVICES=0,1,2,3 \

examples/transformer_tts/synthesis.py

@@ -14,7 +14,7 @@ from pprint import pprint
 from collections import OrderedDict
 from parakeet import audio
 from parakeet.models.transformer_tts.vocoder import Vocoder
-from parakeet.models.transformer_tts.transformerTTS import TransformerTTS
+from parakeet.models.transformer_tts.transformer_tts import TransformerTTS
 
 def load_checkpoint(step, model_path):
     model_dict, _ = fluid.dygraph.load_dygraph(os.path.join(model_path, step))

examples/transformer_tts/train_transformer.py

@@ -13,8 +13,8 @@ import paddle.fluid as fluid
 import paddle.fluid.dygraph as dg
 import paddle.fluid.layers as layers
 from parakeet.modules.utils import cross_entropy
-from parakeet.models.dataloader.ljspeech import LJSpeechLoader
-from parakeet.models.transformer_tts.transformerTTS import TransformerTTS
+from data import LJSpeechLoader
+from parakeet.models.transformer_tts.transformer_tts import TransformerTTS
 
 def load_checkpoint(step, model_path):
     model_dict, opti_dict = fluid.dygraph.load_dygraph(os.path.join(model_path, step))

examples/transformer_tts/train_vocoder.py

@@ -10,7 +10,7 @@ from pprint import pprint
 import paddle.fluid as fluid
 import paddle.fluid.dygraph as dg
 import paddle.fluid.layers as layers
-from parakeet.models.dataloader.ljspeech import LJSpeechLoader
+from data import LJSpeechLoader
 from parakeet.models.transformer_tts.vocoder import Vocoder
 
 def load_checkpoint(step, model_path):

parakeet/models/fastspeech/decoder.py

 import paddle.fluid.dygraph as dg
 import paddle.fluid as fluid
 from parakeet.modules.utils import *
-from parakeet.models.fastspeech.FFTBlock import FFTBlock
+from parakeet.models.fastspeech.fft_block import FFTBlock
 
 class Decoder(dg.Layer):
     def __init__(self,

parakeet/models/fastspeech/encoder.py

 import paddle.fluid.dygraph as dg
 import paddle.fluid as fluid
 from parakeet.modules.utils import *
-from parakeet.models.fastspeech.FFTBlock import FFTBlock
+from parakeet.models.fastspeech.fft_block import FFTBlock
 
 class Encoder(dg.Layer):
     def __init__(self,

parakeet/models/fastspeech/fastspeech.py

@@ -3,7 +3,7 @@ import paddle.fluid.dygraph as dg
 import paddle.fluid as fluid
 from parakeet.g2p.text.symbols import symbols
 from parakeet.models.transformer_tts.post_convnet import PostConvNet
-from parakeet.models.fastspeech.LengthRegulator import LengthRegulator
+from parakeet.models.fastspeech.length_regulator import LengthRegulator
 from parakeet.models.fastspeech.encoder import Encoder
 from parakeet.models.fastspeech.decoder import Decoder
 

parakeet/models/transformer_tts/transformerTTS.py → parakeet/models/transformer_tts/transformer_tts.py

@@ -11,16 +11,9 @@ class TransformerTTS(dg.Layer):
         self.config = config
 
     def forward(self, characters, mel_input, pos_text, pos_mel):
-        # key (batch_size, seq_len, channel)
-        # c_mask (batch_size, seq_len)
-        # attns_enc (channel / 2, seq_len, seq_len)
-
+        
         key, c_mask, attns_enc = self.encoder(characters, pos_text)
         
-        # mel_output/postnet_output (batch_size, mel_len, n_mel)
-        # attn_probs (128, mel_len, seq_len)
-        # stop_preds (batch_size, mel_len, 1)
-        # attns_dec (128, mel_len, mel_len)
         mel_output, postnet_output, attn_probs, stop_preds, attns_dec = self.decoder(key, key, mel_input, c_mask, pos_mel)
 
         return mel_output, postnet_output, attn_probs, stop_preds, attns_enc, attns_dec

parakeet/models/transformer_tts/vocoder.py

@@ -2,7 +2,7 @@ import paddle.fluid.dygraph as dg
 import paddle.fluid as fluid
 from parakeet.modules.customized import Conv1D
 from parakeet.modules.utils import *
-from parakeet.models.transformer_tts.CBHG import CBHG
+from parakeet.models.transformer_tts.cbhg import CBHG
 
 class Vocoder(dg.Layer):
     """