Fabrice-TIERCELIN commited on
Commit
70a8aa2
β€’
1 Parent(s): 336a1d0
Files changed (1) hide show
  1. app.py +2 -5
app.py CHANGED
@@ -58,9 +58,6 @@ class Tango:
58
  def generate(self, prompt, steps = 100, guidance = 3, samples = 1, disable_progress = True):
59
  # Generate audio for a single prompt string
60
  with torch.no_grad():
61
- print(self.model.inference.__code__.co_varnames)
62
- print(self.vae.decode_first_stage.__code__.co_varnames)
63
- print(self.vae.decode_to_waveform.__code__.co_varnames)
64
  latents = self.model.inference([prompt], self.scheduler, steps, guidance, samples, disable_progress = disable_progress)
65
  mel = self.vae.decode_first_stage(latents)
66
  wave = self.vae.decode_to_waveform(mel)
@@ -163,7 +160,7 @@ with gr.Blocks() as interface:
163
  <li>If you need to generate <b>music</b>, I recommend to use <i>MusicGen</i>,</li>
164
  </ul>
165
  <br/>
166
- 🐌 Slow process... ~2 hours. Your computer must <b><u>not</u></b> enter into standby mode.<br/>You can duplicate this space on a free account, it works on CPU.<br/>
167
  <a href='https://huggingface.co/spaces/Fabrice-TIERCELIN/Text-to-Audio?duplicate=true&hidden=public&hidden=public'><img src='https://img.shields.io/badge/-Duplicate%20Space-blue?labelColor=white&style=flat&logo=data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAABAAAAAQCAYAAAAf8/9hAAAAAXNSR0IArs4c6QAAAP5JREFUOE+lk7FqAkEURY+ltunEgFXS2sZGIbXfEPdLlnxJyDdYB62sbbUKpLbVNhyYFzbrrA74YJlh9r079973psed0cvUD4A+4HoCjsA85X0Dfn/RBLBgBDxnQPfAEJgBY+A9gALA4tcbamSzS4xq4FOQAJgCDwV2CPKV8tZAJcAjMMkUe1vX+U+SMhfAJEHasQIWmXNN3abzDwHUrgcRGmYcgKe0bxrblHEB4E/pndMazNpSZGcsZdBlYJcEL9Afo75molJyM2FxmPgmgPqlWNLGfwZGG6UiyEvLzHYDmoPkDDiNm9JR9uboiONcBXrpY1qmgs21x1QwyZcpvxt9NS09PlsPAAAAAElFTkSuQmCC&logoWidth=14'></a>
168
  <br/>
169
  βš–οΈ You can use, modify and share the generated sounds but not for commercial uses.
@@ -172,7 +169,7 @@ with gr.Blocks() as interface:
172
  input_text = gr.Textbox(label = "Prompt", value = "Snort of a horse", lines = 2, autofocus = True)
173
  with gr.Accordion("Advanced options", open = False):
174
  output_format = gr.Radio(label = "Output format", info = "The file you can dowload", choices = ["mp3", "wav"], value = "wav")
175
- output_number = gr.Slider(label = "Number of generations", info = "1, 2 or 3 output files", minimum = 1, maximum = 3, value = 3, step = 1, interactive = True)
176
  denoising_steps = gr.Slider(label = "Steps", info = "lower=faster & variant, higher=audio quality & similar", minimum = 10, maximum = 200, value = 10, step = 1, interactive = True)
177
  guidance_scale = gr.Slider(label = "Guidance Scale", info = "lower=audio quality, higher=follow the prompt", minimum = 1, maximum = 10, value = 3, step = 0.1, interactive = True)
178
  randomize_seed = gr.Checkbox(label = "\U0001F3B2 Randomize seed", value = True, info = "If checked, result is always different")
 
58
  def generate(self, prompt, steps = 100, guidance = 3, samples = 1, disable_progress = True):
59
  # Generate audio for a single prompt string
60
  with torch.no_grad():
 
 
 
61
  latents = self.model.inference([prompt], self.scheduler, steps, guidance, samples, disable_progress = disable_progress)
62
  mel = self.vae.decode_first_stage(latents)
63
  wave = self.vae.decode_to_waveform(mel)
 
160
  <li>If you need to generate <b>music</b>, I recommend to use <i>MusicGen</i>,</li>
161
  </ul>
162
  <br/>
163
+ 🐌 Slow process... ~5 min. Your computer must <b><u>not</u></b> enter into standby mode.<br/>You can duplicate this space on a free account, it works on CPU.<br/>
164
  <a href='https://huggingface.co/spaces/Fabrice-TIERCELIN/Text-to-Audio?duplicate=true&hidden=public&hidden=public'><img src='https://img.shields.io/badge/-Duplicate%20Space-blue?labelColor=white&style=flat&logo=data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAABAAAAAQCAYAAAAf8/9hAAAAAXNSR0IArs4c6QAAAP5JREFUOE+lk7FqAkEURY+ltunEgFXS2sZGIbXfEPdLlnxJyDdYB62sbbUKpLbVNhyYFzbrrA74YJlh9r079973psed0cvUD4A+4HoCjsA85X0Dfn/RBLBgBDxnQPfAEJgBY+A9gALA4tcbamSzS4xq4FOQAJgCDwV2CPKV8tZAJcAjMMkUe1vX+U+SMhfAJEHasQIWmXNN3abzDwHUrgcRGmYcgKe0bxrblHEB4E/pndMazNpSZGcsZdBlYJcEL9Afo75molJyM2FxmPgmgPqlWNLGfwZGG6UiyEvLzHYDmoPkDDiNm9JR9uboiONcBXrpY1qmgs21x1QwyZcpvxt9NS09PlsPAAAAAElFTkSuQmCC&logoWidth=14'></a>
165
  <br/>
166
  βš–οΈ You can use, modify and share the generated sounds but not for commercial uses.
 
169
  input_text = gr.Textbox(label = "Prompt", value = "Snort of a horse", lines = 2, autofocus = True)
170
  with gr.Accordion("Advanced options", open = False):
171
  output_format = gr.Radio(label = "Output format", info = "The file you can dowload", choices = ["mp3", "wav"], value = "wav")
172
+ output_number = gr.Slider(label = "Number of generations", info = "1, 2 or 3 output files", minimum = 1, maximum = 3, value = 1, step = 1, interactive = True)
173
  denoising_steps = gr.Slider(label = "Steps", info = "lower=faster & variant, higher=audio quality & similar", minimum = 10, maximum = 200, value = 10, step = 1, interactive = True)
174
  guidance_scale = gr.Slider(label = "Guidance Scale", info = "lower=audio quality, higher=follow the prompt", minimum = 1, maximum = 10, value = 3, step = 0.1, interactive = True)
175
  randomize_seed = gr.Checkbox(label = "\U0001F3B2 Randomize seed", value = True, info = "If checked, result is always different")