LT4Ryan commited on
Commit
77fe306
·
verified ·
1 Parent(s): 150ffdc

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +12 -11
app.py CHANGED
@@ -123,8 +123,8 @@ def get_transcripts_and_raw_times(audio_path):
123
 
124
  model.cpu()
125
  gc.collect()
126
- #if device == "cuda":
127
- # torch.cuda.empty_cache()
128
 
129
  # Generate CSV
130
  button_update = gr.DownloadButton(visible=False)
@@ -189,17 +189,18 @@ def play_segment(evt: gr.SelectData, raw_ts_list, current_audio_path):
189
  return gr.Audio(value=None, label="Selected Segment")
190
 
191
  article = (
192
- "<p style='font-size: 1.1em;'>"
193
- "AudioDog uses <code><a href='https://huggingface.co/nvidia/parakeet-tdt-0.6b-v2'>parakeet-tdt-0.6b-v2</a></code>, a 600-million-parameter model designed for high-quality English speech recognition."
194
- "</p>"
195
  "<p><strong style='color: red; font-size: 1.2em;'>Key Features:</strong></p>"
196
- "<ul style='font-size: 1.1em;'>"
197
- " <li>Automatic punctuation and capitalization</li>"
198
- " <li>Accurate word-level timestamps (click on a segment in the table below to play it!)</li>"
199
- " <li>Efficiently transcribes long audio segments by chunking them into smaller segments and stitching them together when done.</li>"
200
- " <li>MP3 support for audio input and output, works well on downloaded YouTube videos.</li>"
201
  "</ul>"
 
202
  )
 
203
  examples = [
204
  ["data/example-yt_saTD1u8PorI.mp3"],
205
  ]
@@ -208,7 +209,7 @@ examples = [
208
  # Define an NVIDIA-inspired theme
209
  nvidia_theme = gr_themes.Default(
210
  primary_hue=gr_themes.Color(
211
- c50="#E6F1D9", # Lightest green
212
  c100="#CEE3B3",
213
  c200="#B5D58C",
214
  c300="#9CC766",
 
123
 
124
  model.cpu()
125
  gc.collect()
126
+ if device == "cuda":
127
+ torch.cuda.empty_cache()
128
 
129
  # Generate CSV
130
  button_update = gr.DownloadButton(visible=False)
 
189
  return gr.Audio(value=None, label="Selected Segment")
190
 
191
  article = (
192
+ "<div style='font-size: 1.1em;'>"
193
+ "<p>AudioDog uses <code><a href='https://huggingface.co/nvidia/parakeet-tdt-0.6b-v2'>parakeet-tdt-0.6b-v2</a></code>, a 600-million-parameter model designed for high-quality English speech recognition.</p>"
 
194
  "<p><strong style='color: red; font-size: 1.2em;'>Key Features:</strong></p>"
195
+ "<ul>"
196
+ "<li>Automatic punctuation and capitalization</li>"
197
+ "<li>Accurate word-level timestamps (click on a segment in the table below to play it!)</li>"
198
+ "<li>Efficiently transcribes long audio segments by chunking them into smaller segments and stitching them together when done.</li>"
199
+ "<li>MP3 support for audio input and output, works well on downloaded YouTube videos.</li>"
200
  "</ul>"
201
+ "</div>"
202
  )
203
+
204
  examples = [
205
  ["data/example-yt_saTD1u8PorI.mp3"],
206
  ]
 
209
  # Define an NVIDIA-inspired theme
210
  nvidia_theme = gr_themes.Default(
211
  primary_hue=gr_themes.Color(
212
+ c50="#E5F1D9", # Lightest green
213
  c100="#CEE3B3",
214
  c200="#B5D58C",
215
  c300="#9CC766",