jieminz commited on
Commit
6c34003
·
verified ·
1 Parent(s): 646ac72

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +2 -46
app.py CHANGED
@@ -1,8 +1,5 @@
1
  import gradio as gr
2
- import librosa
3
  from asr import transcribe, ASR_EXAMPLES, ASR_LANGUAGES, ASR_NOTE
4
- from tts import synthesize, TTS_EXAMPLES, TTS_LANGUAGES
5
- from lid import identify, LID_EXAMPLES
6
 
7
 
8
 
@@ -27,44 +24,6 @@ mms_transcribe = gr.Interface(
27
  allow_flagging="never",
28
  )
29
 
30
- mms_synthesize = gr.Interface(
31
- fn=synthesize,
32
- inputs=[
33
- gr.Text(label="Input text"),
34
- gr.Dropdown(
35
- [f"{k} ({v})" for k, v in TTS_LANGUAGES.items()],
36
- label="Language",
37
- value="eng English",
38
- ),
39
- gr.Slider(minimum=0.1, maximum=4.0, value=1.0, step=0.1, label="Speed"),
40
- ],
41
- outputs=[
42
- gr.Audio(label="Generated Audio", type="numpy"),
43
- gr.Text(label="Filtered text after removing OOVs"),
44
- ],
45
- examples=TTS_EXAMPLES,
46
- title="Text-to-speech",
47
- description=("Generate audio in your desired language from input text."),
48
- allow_flagging="never",
49
- )
50
-
51
- mms_identify = gr.Interface(
52
- fn=identify,
53
- inputs=[
54
- gr.Audio(),
55
- ],
56
- outputs=gr.Label(num_top_classes=10),
57
- examples=LID_EXAMPLES,
58
- title="Language Identification",
59
- description=("Identity the language of input audio."),
60
- allow_flagging="never",
61
- )
62
-
63
- tabbed_interface = gr.TabbedInterface(
64
- [mms_transcribe, mms_synthesize, mms_identify],
65
- ["Speech-to-text", "Text-to-speech", "Language Identification"],
66
- )
67
-
68
  with gr.Blocks() as demo:
69
  gr.HTML(
70
  """
@@ -79,16 +38,13 @@ with gr.Blocks() as demo:
79
  "<p align='center' style='font-size: 20px;'>MMS: Scaling Speech Technology to 1000+ languages demo. See our <a href='https://ai.facebook.com/blog/multilingual-model-speech-recognition/'>blog post</a> and <a href='https://arxiv.org/abs/2305.13516'>paper</a>.</p>"
80
  )
81
  gr.HTML(
82
- """<center>Click on the appropriate tab to explore Speech-to-text (ASR), Text-to-speech (TTS) and Language identification (LID) demos. </center>"""
83
- )
84
- gr.HTML(
85
- """<center>You can also finetune MMS models on your data using the recipes provides here - <a href='https://huggingface.co/blog/mms_adapters'>ASR</a> <a href='https://github.com/ylacombe/finetune-hf-vits'>TTS</a> </center>"""
86
  )
87
  gr.HTML(
88
  """<center><a href="https://huggingface.co/spaces/facebook/MMS?duplicate=true" style="display: inline-block;margin-top: .5em;margin-right: .25em;" target="_blank"><img style="margin-bottom: 0em;display: inline;margin-top: -.25em;" src="https://bit.ly/3gLdBN6" alt="Duplicate Space"></a> for more control and no queue.</center>"""
89
  )
90
 
91
- tabbed_interface.render()
92
  gr.HTML(
93
  """
94
  <div class="footer" style="text-align:center">
 
1
  import gradio as gr
 
2
  from asr import transcribe, ASR_EXAMPLES, ASR_LANGUAGES, ASR_NOTE
 
 
3
 
4
 
5
 
 
24
  allow_flagging="never",
25
  )
26
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
27
  with gr.Blocks() as demo:
28
  gr.HTML(
29
  """
 
38
  "<p align='center' style='font-size: 20px;'>MMS: Scaling Speech Technology to 1000+ languages demo. See our <a href='https://ai.facebook.com/blog/multilingual-model-speech-recognition/'>blog post</a> and <a href='https://arxiv.org/abs/2305.13516'>paper</a>.</p>"
39
  )
40
  gr.HTML(
41
+ """<center>You can also finetune MMS models on your data using the recipes provides <a href='https://huggingface.co/blog/mms_adapters'>here</a>.</center>"""
 
 
 
42
  )
43
  gr.HTML(
44
  """<center><a href="https://huggingface.co/spaces/facebook/MMS?duplicate=true" style="display: inline-block;margin-top: .5em;margin-right: .25em;" target="_blank"><img style="margin-bottom: 0em;display: inline;margin-top: -.25em;" src="https://bit.ly/3gLdBN6" alt="Duplicate Space"></a> for more control and no queue.</center>"""
45
  )
46
 
47
+ mms_transcribe.render()
48
  gr.HTML(
49
  """
50
  <div class="footer" style="text-align:center">