mizoru commited on
Commit
42d4ff2
1 Parent(s): 7c7fe0a

reverting to prior to add version choice

Browse files
Files changed (1) hide show
  1. app.py +6 -24
app.py CHANGED
@@ -11,25 +11,14 @@ def get_x(df):
11
  def get_y(df):
12
  return df.pattern
13
 
14
- learn_removeSilence = load_learner('xresnet50_pitch3_removeSilence.pkl')
15
 
16
- learn_plain = load_learner('xresnet50_pitch3.pkl')
17
 
18
- labels = learn_removeSilence.dls.vocab
19
-
20
- #def process(Record, Upload, version):
21
- # if version == 'remove silence':
22
- # return predict(Record, Upload, learn_removeSilence)
23
- # elif version == 'plain':
24
- # return predict(Record, Upload, learn_plain)
25
-
26
- def predict(Record, Upload, version):
27
  if Upload: path = Upload
28
  else: path = Record
29
- if version == 'remove silence':
30
- spec,pred,pred_idx,probs = learn_removeSilence.predict(str(path), with_input=True)
31
- elif version == 'plain':
32
- spec,pred,pred_idx,probs = learn_plain.predict(str(path), with_input=True)
33
  fig,ax = plt.subplots(figsize=(16,10))
34
  show_image(spec, ax=ax)
35
  ax.invert_yaxis()
@@ -42,16 +31,9 @@ description = "This model will predict the pitch accent pattern of a word based
42
 
43
  article="<p style='text-align: center'><a href='https://mizoru.github.io/blog/2021/12/25/Japanese-pitch.html' target='_blank'>How did I make this and what is it for?</a></p>"
44
 
45
- ex_paths = ['代わる.mp3'],['大丈夫な.mp3'],['熱くない.mp3'], ['あめー雨.mp3'], ['あめー飴.mp3']
46
-
47
- examples = [[path, '', 'remove silence'] for path in ex_paths]
48
 
49
  enable_queue=True
50
 
51
- gr.Interface(fn=predict,
52
- inputs=[gr.inputs.Audio(source='microphone', type='filepath', optional=True),
53
- gr.inputs.Audio(source='upload', type='filepath', optional=True),
54
- gr.inputs.Radio(choices=['plain','remove silence'], type="value", default='remove silence', label='version')
55
- ],
56
- outputs= [gr.outputs.Label(num_top_classes=3), gr.outputs.Image(type="plot", label='Spectrogram')], title=title,description=description,article=article,examples=examples).launch(debug=True, enable_queue=enable_queue)
57
 
 
11
  def get_y(df):
12
  return df.pattern
13
 
14
+ learn = load_learner('xresnet50_pitch3_removeSilence.pkl')
15
 
16
+ labels = learn.dls.vocab
17
 
18
+ def predict(Record, Upload):
 
 
 
 
 
 
 
 
19
  if Upload: path = Upload
20
  else: path = Record
21
+ spec,pred,pred_idx,probs = learn.predict(str(path), with_input=True)
 
 
 
22
  fig,ax = plt.subplots(figsize=(16,10))
23
  show_image(spec, ax=ax)
24
  ax.invert_yaxis()
 
31
 
32
  article="<p style='text-align: center'><a href='https://mizoru.github.io/blog/2021/12/25/Japanese-pitch.html' target='_blank'>How did I make this and what is it for?</a></p>"
33
 
34
+ examples = [['代わる.mp3'],['大丈夫な.mp3'],['熱くない.mp3'], ['あめー雨.mp3'], ['あめー飴.mp3']]
 
 
35
 
36
  enable_queue=True
37
 
38
+ gr.Interface(fn=predict,inputs=[gr.inputs.Audio(source='microphone', type='filepath', optional=True), gr.inputs.Audio(source='upload', type='filepath', optional=True)], outputs= [gr.outputs.Label(num_top_classes=3), gr.outputs.Image(type="plot", label='Spectrogram')], title=title,description=description,article=article,examples=examples).launch(debug=True, enable_queue=enable_queue)
 
 
 
 
 
39