Pendrokar commited on
Commit
015adff
โ€ข
1 Parent(s): a0a523d

arpabet durations

Browse files
Files changed (1) hide show
  1. gr_client.py +17 -5
gr_client.py CHANGED
@@ -114,12 +114,24 @@ def predict(
114
  deepmoji_checked, # bool
115
  api_name="/predict"
116
  )
 
117
  json_data = json.loads(response)
118
- arpabet = json_data['arpabet'].replace('|<PAD>|', ' ')
 
 
 
 
 
 
 
 
 
 
 
119
 
120
  return [
121
  wav_path,
122
- arpabet,
123
  round(json_data['em_angry'][0], 2),
124
  round(json_data['em_happy'][0], 2),
125
  round(json_data['em_sad'][0], 2),
@@ -212,7 +224,7 @@ language_radio = gr.Radio(
212
  info="Will be more monotone and have an English accent. Tested mostly by a native Briton."
213
  )
214
 
215
- with gr.Blocks() as demo:
216
  gr.Markdown("# xVASynth TTS")
217
 
218
  with gr.Row(): # Main row for inputs and language selection
@@ -248,7 +260,7 @@ with gr.Blocks() as demo:
248
  with gr.Column(): # Input column
249
  happy_slider = gr.Slider(0, 1.0, value=0, step=0.05, label="๐Ÿ˜ƒ Happiness", info="Tread lightly beyond 0.7")
250
  surprise_slider = gr.Slider(0, 1.0, value=0, step=0.05, label="๐Ÿ˜ฎ Surprise", info="Can oversaturate Happiness")
251
- deepmoji_checkbox = gr.Checkbox(label="Use DeepMoji", info="Auto adjust emotional values")
252
 
253
  # Event handling using click
254
  btn = gr.Button("Generate")
@@ -257,7 +269,7 @@ with gr.Blocks() as demo:
257
  with gr.Column(): # Input column
258
  output_wav = gr.Audio(label="22kHz audio output", type="filepath", editable=False)
259
  with gr.Column(): # Input column
260
- output_arpabet = gr.Textbox(label="ARPAbet", interactive=False)
261
 
262
  btn.click(
263
  fn=predict,
 
114
  deepmoji_checked, # bool
115
  api_name="/predict"
116
  )
117
+
118
  json_data = json.loads(response)
119
+
120
+ arpabet_html = '<h6>ARPAbet & Durations</h6>'
121
+ arpabet_symbols = json_data['arpabet'].split('|')
122
+ for symb_i in range(len(json_data['durations'])):
123
+ if (arpabet_symbols[symb_i] == '<PAD>'):
124
+ continue
125
+
126
+ arpabet_html += '<strong class="arpabet" style="padding: 0 '\
127
+ + str(round(float(json_data['durations'][symb_i]/2), 1))\
128
+ +'em">'\
129
+ + arpabet_symbols[symb_i]\
130
+ + '</strong> '
131
 
132
  return [
133
  wav_path,
134
+ arpabet_html,
135
  round(json_data['em_angry'][0], 2),
136
  round(json_data['em_happy'][0], 2),
137
  round(json_data['em_sad'][0], 2),
 
224
  info="Will be more monotone and have an English accent. Tested mostly by a native Briton."
225
  )
226
 
227
+ with gr.Blocks(css=".arpabet {display: inline-block; background-color: gray; border-radius: 5px; font-size: 120%; margin: 0.1em 0}") as demo:
228
  gr.Markdown("# xVASynth TTS")
229
 
230
  with gr.Row(): # Main row for inputs and language selection
 
260
  with gr.Column(): # Input column
261
  happy_slider = gr.Slider(0, 1.0, value=0, step=0.05, label="๐Ÿ˜ƒ Happiness", info="Tread lightly beyond 0.7")
262
  surprise_slider = gr.Slider(0, 1.0, value=0, step=0.05, label="๐Ÿ˜ฎ Surprise", info="Can oversaturate Happiness")
263
+ deepmoji_checkbox = gr.Checkbox(label="Use DeepMoji", info="Auto adjust emotional values", value=True)
264
 
265
  # Event handling using click
266
  btn = gr.Button("Generate")
 
269
  with gr.Column(): # Input column
270
  output_wav = gr.Audio(label="22kHz audio output", type="filepath", editable=False)
271
  with gr.Column(): # Input column
272
+ output_arpabet = gr.HTML(label="ARPAbet")
273
 
274
  btn.click(
275
  fn=predict,