Update app.py

app.py

@@ -1,12 +1,20 @@
+# Imports
+import gradio as gr
 import spaces
+
 from kokoro import KModel, KPipeline
-import gradio as gr
 
-CHAR_LIMIT = 5000
+# Pre-Initialize
+DEVICE = "auto"
+if DEVICE == "auto":
+    DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
+print(f"[SYSTEM] | Using {DEVICE} type compute device.")
 
-PIPELINES = {lang: KPipeline(lang_code=lang, model=False) for lang in "ab"}
-PIPELINES["a"].g2p.lexicon.golds["kokoro"] = "kˈOkəɹO"
-PIPELINES["b"].g2p.lexicon.golds["kokoro"] = "kˈQkəɹQ"
+# Variables
+CHAR_LIMIT = 2000
+
+DEFAULT_INPUT = ""
+DEFAULT_VOICE = "af_heart"
 
 CHOICES = {
     "🇺🇸 🚺 Heart ❤️": "af_heart",
@@ -39,45 +47,55 @@ CHOICES = {
     "🇬🇧 🚹 Daniel": "bm_daniel",
 }
 
+PIPELINES = {lang: KPipeline(lang_code=lang, model=False) for lang in "ab"}
+PIPELINES["a"].g2p.lexicon.golds["kokoro"] = "kˈOkəɹO"
+PIPELINES["b"].g2p.lexicon.golds["kokoro"] = "kˈQkəɹQ"
+
 for v in CHOICES.values():
     PIPELINES[v[0]].load_voice(v)
 
 MODEL = KModel().eval()
 
-def generate_first(text, voice="af_heart", speed=1):
-    text = text.strip()[:CHAR_LIMIT]
+css = '''
+.gradio-container{max-width: 560px !important}
+h1{text-align:center}
+footer {
+    visibility: hidden
+}
+'''
+
+# Functions
+def generate(text=DEFAULT_INPUT, voice=DEFAULT_VOICE, speed=1):
+    text = text.strip()[:CHAR_LIMIT] + "."
     pipeline = PIPELINES[voice[0]]
     pack = pipeline.load_voice(voice)
+    
     for _, ps, _ in pipeline(text, voice, speed):
         ref_s = pack[len(ps) - 1]
         audio = MODEL(ps, ref_s, speed)
         return (24000, audio.numpy())
-    return None
 
-def predict(text, voice="af_heart", speed=1):
-    return generate_first(text, voice, speed)
-
-def generate_all(text, voice="af_heart", speed=1):
-    text = text.strip()[:CHAR_LIMIT]
-    pipeline = PIPELINES[voice[0]]
-    pack = pipeline.load_voice(voice)
-    for _, ps, _ in pipeline(text, voice, speed):
-        ref_s = pack[len(ps) - 1]
-        audio = MODEL(ps, ref_s, speed)
-        yield 24000, audio.numpy()
+def cloud():
+    print("[CLOUD] | Space maintained.")
 
 @spaces.GPU()
 def gpu():
     return
 
-with gr.Blocks() as app:
-    with gr.Row():
-        text_input = gr.Textbox(label="input text")
-        voice_input = gr.Dropdown(list(CHOICES.items()), value="af_heart", label="voice")
-        speed_input = gr.Slider(minimum=0.5, maximum=2, value=1, step=0.1, label="speed")
-    out_audio = gr.Audio(label="output audio", interactive=False, autoplay=True)
-    gen_btn = gr.Button("generate")
-    gen_btn.click(fn=generate_first, inputs=[text_input, voice_input, speed_input], outputs=out_audio)
+# Initialize
+with gr.Blocks(css=css) as main:
+    with gr.Column():
+        input = gr.Textbox(lines=1, value=DEFAULT_INPUT, label="Input")
+        voice_input = gr.Dropdown(list(CHOICES.items()), value=DEFAULT_VOICE, label="Voice")
+        speed_input = gr.Slider(minimum=0.5, maximum=2, value=1, step=0.1, label="Speed")
+        
+        submit = gr.Button("▶")
+        maintain = gr.Button("☁️")
+        
+    with gr.Column():
+        output = gr.Audio(label="Output")
+    
+    submit.click(fn=generate, inputs=[input, voice_input, speed_input], outputs=output)
+    maintain.click(cloud, inputs=[], outputs=[], queue=False)
 
-if __name__ == "__main__":
-    app.launch()
+main.launch(show_api=True)