Update app.py

app.py

@@ -38,7 +38,7 @@ def get_text(text, hps):
 
 
 hps = utils.get_hparams_from_file("configs/ljs_base.json")
-hps_ms = utils.get_hparams_from_file("configs/vctk_base.json")
+hps_ms = utils.get_hparams_from_file("configs/config.json")
 net_g_ms = SynthesizerTrn(
     len(symbols),
     hps_ms.data.filter_length // 2 + 1,
@@ -50,7 +50,7 @@ import numpy as np
 
 hubert = torch.hub.load("bshall/hubert:main", "hubert_soft")
 
-_ = utils.load_checkpoint("G_312000.pth", net_g_ms, None)
+_ = utils.load_checkpoint("G_18000.pth", net_g_ms, None)
 
 def vc_fn(input_audio,vc_transform):
     if input_audio is None:
@@ -77,7 +77,7 @@ def vc_fn(input_audio,vc_transform):
         print(sampling_rate)
         f0 = resize2d(f0, len(soft[:, 0])) * vc_transform
         soft[:, 0] = f0 / 10
-    sid = torch.LongTensor([1])
+    sid = torch.LongTensor([int(vc_id)])
     stn_tst = torch.FloatTensor(soft)
     with torch.no_grad():
         x_tst = stn_tst.unsqueeze(0)
@@ -95,6 +95,7 @@ with app:
         with gr.TabItem("Basic"):
             vc_input3 = gr.Audio(label="Input Audio (30s limitation)")
             vc_transform = gr.Number(label="transform",value=1.0)
+            vc_id=gr.Textbox(lines=1, placeholder="speaker_id"),
             vc_submit = gr.Button("Convert", variant="primary")
             vc_output1 = gr.Textbox(label="Output Message")
             vc_output2 = gr.Audio(label="Output Audio")