add max_iteration into configuration, fix np.pad for lower versions of numpy

examples/deepvoice3/README.md

@@ -87,7 +87,7 @@ runs/Jul07_09-39-34_instance-mqcyj27y-4/
 ...
 ```
 
-Since e use waveflow to synthesize audio while training, so download the trained waveflow model and extract it in current directory before training.
+Since we use waveflow to synthesize audio while training, so download the trained waveflow model and extract it in current directory before training.
 
 ```bash
 wget https://paddlespeech.bj.bcebos.com/Parakeet/waveflow_res128_ljspeech_ckpt_1.0.zip

examples/deepvoice3/configs/ljspeech.yaml

@@ -39,6 +39,7 @@ clip_value: 5.0
 clip_norm: 100.0
 
 # training:
+max_iteration: 1000000
 batch_size: 16
 report_interval: 10000
 save_interval: 10000

examples/deepvoice3/data.py

@@ -62,10 +62,8 @@ class DataCollector(object):
         for example in examples:
             text, spec, mel, _ = example
             text_seqs.append(en.text_to_sequence(text, self.p_pronunciation))
-            # if max_frames - mel.shape[0] < 0:
-            #     import pdb; pdb.set_trace()
-            specs.append(np.pad(spec, [(0, max_frames - spec.shape[0]), (0, 0)]))
-            mels.append(np.pad(mel, [(0, max_frames - mel.shape[0]), (0, 0)]))
+            specs.append(np.pad(spec, [(0, max_frames - spec.shape[0]), (0, 0)], mode="constant"))
+            mels.append(np.pad(mel, [(0, max_frames - mel.shape[0]), (0, 0)], mode="constant"))
 
         specs = np.stack(specs)
         mels = np.stack(mels)

examples/deepvoice3/train.py

@@ -81,7 +81,7 @@ def train(args, config):
     optim = create_optimizer(model, config)
 
     global global_step
-    max_iteration = 1000000
+    max_iteration = config["max_iteration"]
     
     iterator = iter(tqdm.tqdm(train_loader))
     while global_step <= max_iteration: