AlexK-PL commited on
Commit
cadb9c5
·
1 Parent(s): eec8d3d

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +4 -16
app.py CHANGED
@@ -10,10 +10,8 @@ from melgan.utils.hparams import load_hparam
10
 
11
  import torch
12
  import numpy as np
13
- import random
14
 
15
  from matplotlib import pyplot as plt
16
- from matplotlib import gridspec
17
 
18
  # Adjust vertical spacing between subplots
19
  plt.subplots_adjust(hspace=0.15) # You can adjust the value as needed
@@ -87,21 +85,11 @@ def synthesize(text, gst_1, gst_2, gst_3):
87
  fig_mel = plot_spec_align(mel_outputs_postnet, alignments)
88
 
89
  return (22050, audio_numpy), fig_mel # fig_align
 
90
 
91
-
92
- def randomize_style_weights():
93
- rand_weight_1 = random.randrange(0.3, 0.45, 0.01)
94
- rand_weight_2 = random.randrange(0.3, 0.45, 0.01)
95
- rand_weight_3 = 1.0 - (rand_weight_1 + rand_weight_2)
96
-
97
- return rand_weight_1, rand_weight_2, rand_weight_3
98
-
99
-
100
- w1, w2, w3 = randomize_style_weights()
101
-
102
- iface = gr.Interface(fn=synthesize, inputs=[gr.Textbox(label="Input Text"), gr.Slider(0.2, 0.45, label="First style token weight:", value=w1),
103
- gr.Slider(0.2, 0.45, label="Second style token weight:", value=w2),
104
- gr.Slider(0.2, 0.45, label="Third style token weight:",value=w3)],
105
  outputs=[gr.Audio(label="Generated Speech", type="numpy"), gr.Plot(label="Output"),],
106
  title="Single-Head Attention Tacotron2 with Style Tokens", description=DESCRIPTION)
107
  iface.launch()
 
10
 
11
  import torch
12
  import numpy as np
 
13
 
14
  from matplotlib import pyplot as plt
 
15
 
16
  # Adjust vertical spacing between subplots
17
  plt.subplots_adjust(hspace=0.15) # You can adjust the value as needed
 
85
  fig_mel = plot_spec_align(mel_outputs_postnet, alignments)
86
 
87
  return (22050, audio_numpy), fig_mel # fig_align
88
+
89
 
90
+ iface = gr.Interface(fn=synthesize, inputs=[gr.Textbox(label="Input Text"), gr.Slider(0.2, 0.45, label="First style token weight:", value=0.4),
91
+ gr.Slider(0.2, 0.45, label="Second style token weight:", value=0.26),
92
+ gr.Slider(0.2, 0.45, label="Third style token weight:", value=0.33)],
 
 
 
 
 
 
 
 
 
 
 
93
  outputs=[gr.Audio(label="Generated Speech", type="numpy"), gr.Plot(label="Output"),],
94
  title="Single-Head Attention Tacotron2 with Style Tokens", description=DESCRIPTION)
95
  iface.launch()