Update app.py

app.py

@@ -92,7 +92,7 @@ CHOICES = {
 '🇺🇸 🚺 Alloy': 'af_alloy',
 '🇺🇸 🚺 Jessica': 'af_jessica',
 '🇺🇸 🚺 River': 'af_river',
-
+    
 '🇺🇸 🚹 Michael': 'am_michael',
 '🇺🇸 🚹 Fenrir': 'am_fenrir',
 '🇺🇸 🚹 Puck': 'am_puck',
@@ -102,46 +102,46 @@ CHOICES = {
 '🇺🇸 🚹 Onyx': 'am_onyx',
 '🇺🇸 🚹 Santa': 'am_santa',
 '🇺🇸 🚹 Adam': 'am_adam',
-
+    
 '🇬🇧 🚺 Emma': 'bf_emma',
 '🇬🇧 🚺 Isabella': 'bf_isabella',
 '🇬🇧 🚺 Alice': 'bf_alice',
 '🇬🇧 🚺 Lily': 'bf_lily',
-
+    
 '🇬🇧 🚹 George': 'bm_george',
 '🇬🇧 🚹 Fable': 'bm_fable',
 '🇬🇧 🚹 Lewis': 'bm_lewis',
 '🇬🇧 🚹 Daniel': 'bm_daniel',
-
+    
 '🇪🇸 🚺 Dora': 'ef_dora',
-
+    
 '🇪🇸 🚹 Alex': 'em_alex',
 '🇪🇸 🚹 Santa': 'em_santa',
-
+    
 '🇫🇷 🚺 Siwis': 'ff_siwis',
-
+    
 '🇮🇳 🚹 Alpha': 'hf_alpha',
 '🇮🇳 🚹 Beta': 'hf_beta',
-
+    
 '🇮🇳 🚹 Omega': 'hm_omega',
 '🇮🇳 🚹 Psi': 'hm_psi',
-
+    
 '🇮🇹 🚺 Sara': 'if_sara',
-
+    
 '🇮🇹 🚺 Nicola': 'im_nicola',
-
+    
 '🇯🇵 🚹 Alpha': 'jf_alpha',
 '🇯🇵 🚹 Gongitsune': 'jf_gongitsune',
 '🇯🇵 🚹 Nezumi': 'jf_nezumi',
 '🇯🇵 🚹 Tebukuro': 'jf_tebukuro',
-
+    
 '🇯🇵 🚹 Kumo': 'jm_kumo',
-
+    
 '🇧🇷 🚺 Dora': 'pf_dora',
-
+    
 '🇧🇷 🚹 Alex': 'pm_alex',
 '🇧🇷 🚹 Santa': 'pm_santa',
-
+    
 '🇨🇳 🚺 Xiaobei': 'zf_xiaobei',
 '🇨🇳 🚺 Xiaoni': 'zf_xiaoni',
 '🇨🇳 🚺 Xiaoxiao': 'zf_xiaoxiao',
@@ -155,12 +155,23 @@ CHOICES = {
 for v in CHOICES.values():
     pipelines[v[0]].load_voice(v)
 
+TOKEN_NOTE = '''
+💡 Customize pronunciation with Markdown link syntax and /slashes/ like [Kokoro](/kˈOkəɹO/)
+💬 To adjust intonation, try punctuation ;:,.!?—…"()“” or stress ˈ and ˌ
+⬇️ Lower stress [1 level](-1) or [2 levels](-2)
+⬆️ Raise stress 1 level [or](+2) 2 levels (only works on less stressed, usually short words)
+'''
+
 with gr.Blocks() as generate_tab:
     out_audio = gr.Audio(label='Output Audio', interactive=False, streaming=False, autoplay=True)
     generate_btn = gr.Button('Generate', variant='primary')
+    with gr.Accordion('Output Tokens', open=True):
+        out_ps = gr.Textbox(interactive=False, show_label=False, info='Tokens used to generate the audio, up to 510 context length.')
+        tokenize_btn = gr.Button('Tokenize', variant='secondary')
+        gr.Markdown(TOKEN_NOTE)
+        predict_btn = gr.Button('Predict', variant='secondary', visible=False)
 
-
-STREAM_NOTE = ['⚠️ There is an unknown Gradio bug that might yield no audio the first time you click `Stream`.']
+STREAM_NOTE = ['⚠️ There is an unknown Gradio bug that might yield no audio the first time you click Stream.']
 if CHAR_LIMIT is not None:
     STREAM_NOTE.append(f'✂️ Each stream is capped at {CHAR_LIMIT} characters.')
     STREAM_NOTE.append('🚀 Want more characters? You can [use Kokoro directly](https://huggingface.co/hexgrad/Kokoro-82M#usage) or duplicate this space:')
@@ -202,9 +213,11 @@ with gr.Blocks() as app:
         with gr.Column():
             gr.TabbedInterface([generate_tab, stream_tab], ['Generate', 'Stream'])
     random_btn.click(fn=get_random_text, inputs=[voice], outputs=[text], api_name=API_NAME)
-    generate_btn.click(fn=generate_first, inputs=[text, voice, speed, use_gpu], outputs=[out_audio], api_name=API_NAME)
+    generate_btn.click(fn=generate_first, inputs=[text, voice, speed, use_gpu], outputs=[out_audio, out_ps], api_name=API_NAME)
+    tokenize_btn.click(fn=tokenize_first, inputs=[text, voice], outputs=[out_ps], api_name=API_NAME)
     stream_event = stream_btn.click(fn=generate_all, inputs=[text, voice, speed, use_gpu], outputs=[out_stream], api_name=API_NAME)
     stop_btn.click(fn=None, cancels=stream_event)
+    predict_btn.click(fn=predict, inputs=[text, voice, speed], outputs=[out_audio], api_name=API_NAME)
 
 if __name__ == '__main__':
-    app.queue(api_open=API_OPEN).launch(show_api=API_OPEN, ssr_mode=True)
+    app.queue(api_open=API_OPEN).launch(show_api=API_OPEN, ssr_mode=True)