select device when synthesize

docs/src/demo.rst

@@ -27,7 +27,10 @@ Analysis/synthesis
 Audio samples generated from ground-truth spectrograms with a vocoder.
 
 .. raw:: html
-
+
+    <b>LJSpeech(English)</b>
+    <br>
+    </br>
     <table>
         <tr>
             <th  align="left"> GT </th>
@@ -101,6 +104,12 @@ Audio samples generated from ground-truth spectrograms with a vocoder.
             </td>
         </tr>
     </table>
+
+    <br>
+    </br>
+    <b>CSMSC(Chinese)</b>
+    <br>
+    </br>
 
     <table>
         <tr>

examples/GANVocoder/parallelwave_gan/baker/synthesize_from_wav.py

@@ -105,6 +105,9 @@ def main():
     parser.add_argument("--verbose", type=int, default=1, help="verbose.")
 
     args = parser.parse_args()
+
+    paddle.set_device(args.device)
+
     config = get_cfg_default()
     if args.config:
         config.merge_from_file(args.config)

examples/fastspeech2/aishell3/synthesize_e2e.py

@@ -149,6 +149,9 @@ def main():
     parser.add_argument("--verbose", type=int, default=1, help="verbose.")
 
     args = parser.parse_args()
+
+    paddle.set_device(args.device)
+
     with open(args.fastspeech2_config) as f:
         fastspeech2_config = CfgNode(yaml.safe_load(f))
     with open(args.pwg_config) as f:

examples/fastspeech2/baker/synthesize_e2e.py

@@ -140,6 +140,9 @@ def main():
     parser.add_argument("--verbose", type=int, default=1, help="verbose.")
 
     args = parser.parse_args()
+
+    paddle.set_device(args.device)
+
     with open(args.fastspeech2_config) as f:
         fastspeech2_config = CfgNode(yaml.safe_load(f))
     with open(args.pwg_config) as f:

examples/fastspeech2/ljspeech/synthesize_e2e.py

@@ -148,6 +148,9 @@ def main():
     parser.add_argument("--verbose", type=int, default=1, help="verbose.")
 
     args = parser.parse_args()
+
+    paddle.set_device(args.device)
+
     with open(args.fastspeech2_config) as f:
         fastspeech2_config = CfgNode(yaml.safe_load(f))
     with open(args.pwg_config) as f:

examples/fastspeech2/synthesize.py

@@ -149,6 +149,9 @@ def main():
     parser.add_argument("--verbose", type=int, default=1, help="verbose.")
 
     args = parser.parse_args()
+
+    paddle.set_device(args.device)
+
     with open(args.fastspeech2_config) as f:
         fastspeech2_config = CfgNode(yaml.safe_load(f))
     with open(args.pwg_config) as f:

examples/speedyspeech/baker/conf/default.yaml

@@ -1,20 +1,20 @@
 ###########################################################
 #                FEATURE EXTRACTION SETTING               #
 ###########################################################
-fs: 24000     # Sampling rate.
-n_fft: 2048           # FFT size.
-n_shift: 300            # Hop size.
-win_length: 1200         # Window length.
-                         # If set to null, it will be the same as fft_size.
-window: "hann"           # Window function.
-n_mels: 80             # Number of mel basis.
-fmin: 80                 # Minimum freq in mel basis calculation.
-fmax: 7600               # Maximum frequency in mel basis calculation.
+fs: 24000           # Sampling rate.
+n_fft: 2048         # FFT size.
+n_shift: 300        # Hop size.
+win_length: 1200    # Window length.
+                    # If set to null, it will be the same as fft_size.
+window: "hann"      # Window function.
+n_mels: 80          # Number of mel basis.
+fmin: 80            # Minimum freq in mel basis calculation.
+fmax: 7600          # Maximum frequency in mel basis calculation.
 
 ###########################################################
 #                       DATA SETTING                      #
 ###########################################################
-batch_size: 256
+batch_size: 32
 num_workers: 4
 
 ###########################################################
@@ -35,13 +35,13 @@ model:
 ###########################################################
 optimizer:
   optim: adam              # optimizer type
-  learning_rate: 0.008     # learning rate
+  learning_rate: 0.001     # learning rate
   max_grad_norm: 5.0
 
 ###########################################################
 #                     TRAINING SETTING                    #
 ###########################################################
-max_epoch: 600
+max_epoch: 300
 num_snapshots: 5
 
 ###########################################################

examples/speedyspeech/baker/synthesize_e2e.py

@@ -175,6 +175,9 @@ def main():
     parser.add_argument("--verbose", type=int, default=1, help="verbose")
 
     args, _ = parser.parse_known_args()
+
+    paddle.set_device(args.device)
+
     with open(args.speedyspeech_config) as f:
         speedyspeech_config = CfgNode(yaml.safe_load(f))
     with open(args.pwg_config) as f:

examples/speedyspeech/synthesize.py

@@ -159,6 +159,9 @@ def main():
     parser.add_argument("--verbose", type=int, default=1, help="verbose")
 
     args, _ = parser.parse_known_args()
+
+    paddle.set_device(args.device)
+
     with open(args.speedyspeech_config) as f:
         speedyspeech_config = CfgNode(yaml.safe_load(f))
     with open(args.pwg_config) as f:

examples/transformer_tts/ljspeech/synthesize_e2e.py

@@ -140,6 +140,9 @@ def main():
     parser.add_argument("--verbose", type=int, default=1, help="verbose.")
 
     args = parser.parse_args()
+
+    paddle.set_device(args.device)
+
     with open(args.transformer_tts_config) as f:
         transformer_tts_config = CfgNode(yaml.safe_load(f))
     with open(args.waveflow_config) as f:

examples/transformer_tts/synthesize.py

@@ -121,6 +121,9 @@ def main():
     parser.add_argument("--verbose", type=int, default=1, help="verbose.")
 
     args = parser.parse_args()
+
+    paddle.set_device(args.device)
+
     with open(args.transformer_tts_config) as f:
         transformer_tts_config = CfgNode(yaml.safe_load(f))
     with open(args.waveflow_config) as f: