Update app.py

app.py

@@ -157,7 +157,7 @@ for v in CHOICES.values():
 
 TOKEN_NOTE = '''
 💡 Customize pronunciation with Markdown link syntax and /slashes/ like `[Kokoro](/kˈOkəɹO/)`
-💬 To adjust intonation, try punctuation `;:,.!?—…"()“”`
+💬 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)
 '''
@@ -165,36 +165,18 @@ TOKEN_NOTE = '''
 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')
-
-    # Commenting out the Output Tokens and Stream Tab sections
-    '''
     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`.']
-    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:')
-    STREAM_NOTE = '\n\n'.join(STREAM_NOTE)
-
-    with gr.Blocks() as stream_tab:
-        out_stream = gr.Audio(label='Output Audio Stream', interactive=False, streaming=True, autoplay=True)
-        with gr.Row():
-            stream_btn = gr.Button('Stream', variant='primary')
-            stop_btn = gr.Button('Stop', variant='stop')
-        with gr.Accordion('Note', open=True):
-            gr.Markdown(STREAM_NOTE)
-            gr.DuplicateButton()
-    '''
-
 BANNER_TEXT = '''
 [***Kokoro*** **is an open-weight TTS model with 82 million parameters.**](https://huggingface.co/hexgrad/Kokoro-82M)
 As of January 31st, 2025, Kokoro was the most-liked [**TTS model**](https://huggingface.co/models?pipeline_tag=text-to-speech&sort=likes) and the most-liked [**TTS space**](https://huggingface.co/spaces?sort=likes&search=tts) on Hugging Face.
 This demo only showcases English, but you can directly use the model to access other languages.
 '''
+
 API_OPEN = os.getenv('SPACE_ID') != 'hexgrad/Kokoro-TTS'
 API_NAME = None if API_OPEN else False
 with gr.Blocks() as app:
@@ -215,11 +197,11 @@ with gr.Blocks() as app:
             speed = gr.Slider(minimum=0.5, maximum=2, value=1, step=0.1, label='Speed')
             random_btn = gr.Button('Random Text', variant='secondary')
         with gr.Column():
-            # Only keeping the Generate tab active
-            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)
-
+            gr.TabbedInterface([generate_tab], ['Generate'])
     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, out_ps], api_name=API_NAME)
+    tokenize_btn.click(fn=tokenize_first, inputs=[text, voice], outputs=[out_ps], api_name=API_NAME)
+    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)