new commits

demo_cli.py

@@ -21,7 +21,7 @@ if __name__ == '__main__':
                         help="weight of input audio for voice filter")
     parser.add_argument("--griffin_lim",
                         action="store_true",
-                        help="if True, use vocoder, else use griffin-lim")
+                        help="if True, use griffin-lim, else use vocoder")
     parser.add_argument("--cpu", action="store_true", help=\
         "If True, processing is done on CPU, even when a GPU is available.")
     parser.add_argument("--no_sound", action="store_true", help=\
@@ -155,9 +155,7 @@ if __name__ == '__main__':
         # get duration info from input audio
         message2 = "Reference voice: enter an audio folder of a voice to be cloned (mp3, " \
                 f"wav, m4a, flac, ...):({i+1}/{num_of_input_audio})\n"
-        # in_fpath = Path(input(message2).replace("\"", "").replace("\'", ""))
-
-        # in_fpath = Path("/home/liuhaozhe/signal_processing_projs/collected_audios/celeb_audios/trimmed/emma_watson_trim.wav")
+        in_fpath = Path(input(message2).replace("\"", "").replace("\'", ""))
 
         fpath_without_ext = os.path.splitext(str(in_fpath))[0]
         speaker_name = os.path.normpath(fpath_without_ext).split(os.sep)[-1]

vocoder/hparams.py

@@ -32,7 +32,7 @@ voc_res_blocks = 10
 
 # Training
 voc_batch_size = 256
-voc_lr = 1e-5
+voc_lr = 1e-6
 voc_gen_at_checkpoint = 5           # number of samples to generate at each checkpoint
 voc_pad = 2                         # this will pad the input so that the resnet can 'see' wider 
                                     # than input length

vocoder_train.py

@@ -30,7 +30,7 @@ if __name__ == "__main__":
         "of those weights and wavs generated during training.")
     parser.add_argument("-g", "--ground_truth", action="store_true", help= \
         "Train on ground truth spectrograms (<datasets_root>/SV2TTS/synthesizer/mels).")
-    parser.add_argument("-s", "--save_every", type=int, default=10000, help= \
+    parser.add_argument("-s", "--save_every", type=int, default=100, help= \
         "Number of steps between updates of the model on the disk. Set to 0 to never save the "
         "model.")
     parser.add_argument("-b", "--backup_every", type=int, default=10000, help= \