Update app.py

app.py

@@ -4,65 +4,86 @@ import asyncio
 import tempfile
 import os
 
-# الدالة الرئيسية التي تستقبل المعاملات الجديدة
+# Map readable labels to Edge TTS internal IDs
+VOICE_MAP = {
+    "رجل (مصري)": "ar-EG-ShakirNeural",
+    "سيدة (مصرية)": "ar-EG-SalmaNeural",
+    "رجل (سعودي)": "ar-SA-HamedNeural",
+    "سيدة (سعودية)": "ar-SA-ZariyahNeural",
+    "English (US) M": "en-US-EricNeural",
+    "English (US) F": "en-US-AriaNeural"
+}
+
 async def text_to_speech_edge(text, voice, emotion, is_symbol, rate, pitch):
-    if not text.strip():
-        return None
+    """
+    Generates speech using Edge TTS.
+    Args:
+        text (str): The text to speak.
+        voice (str): Voice ID or Label.
+        emotion (str): Ignored (kept for compatibility).
+        is_symbol (bool): Ignored (kept for compatibility).
+        rate (str): Speed change (e.g., "+0%", "+20%").
+        pitch (str): Pitch change (e.g., "+0Hz", "+2st").
+    """
     
-    # التعامل مع القيم الافتراضية إذا كانت فارغة
-    if not rate: rate = "+0%"
-    if not pitch: pitch = "+0Hz"
-    
-    # تحديد اسم الصوت بناءً على المدخل
-    voice_map = {
-        "رجل (مصري)": "ar-EG-ShakirNeural",
-        "سيدة (مصرية)": "ar-EG-SalmaNeural",
-        "رجل (سعودي)": "ar-SA-HamedNeural",
-        "سيدة (سعودية)": "ar-SA-ZariyahNeural",
-        "English (US) M": "en-US-EricNeural",
-        "English (US) F": "en-US-AriaNeural"
-    }
+    # 1. Validation
+    if not text or not text.strip():
+        return None
+
+    # 2. Voice Resolution
+    # Check if the input is a known Key (Label) or Value (ID)
+    selected_voice = "ar-SA-HamedNeural" # Default
+    if voice in VOICE_MAP:
+        selected_voice = VOICE_MAP[voice]
+    elif voice in VOICE_MAP.values():
+        selected_voice = voice
     
-    # استخدام المفتاح المباشر إذا لم يكن في القاموس
-    selected_voice = voice_map.get(voice, voice)
+    # 3. Parameter Sanitization
+    # EdgeTTS crashes if rate/pitch are None or empty strings.
+    # We ensure they always have a valid default value.
+    final_rate = rate if rate and isinstance(rate, str) and len(rate.strip()) > 0 else "+0%"
+    final_pitch = pitch if pitch and isinstance(pitch, str) and len(pitch.strip()) > 0 else "+0Hz"
     
+    print(f"Generating: TextLen={len(text)}, Voice={selected_voice}, Rate={final_rate}, Pitch={final_pitch}")
+
+    # 4. Temp File Creation
+    # delete=False is required for Gradio to serve the file after the function returns
     output_file = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False)
     output_path = output_file.name
     output_file.close()
-    
-    # توليد الصوت مع المعاملات الإضافية
-    communicate = edge_tts.Communicate(text, selected_voice, rate=rate, pitch=pitch)
-    await communicate.save(output_path)
-    
-    return output_path
 
-# بناء الواجهة
-with gr.Blocks() as demo:
-    with gr.Row():
-        text_input = gr.Textbox(label="Text")
-        voice_input = gr.Dropdown(
-            ["رجل (مصري)", "سيدة (مصرية)", "رجل (سعودي)", "سيدة (سعودية)", "English (US) M", "English (US) F"], 
-            label="Voice"
-        )
+    try:
+        # 5. Generation
+        communicate = edge_tts.Communicate(text, selected_voice, rate=final_rate, pitch=final_pitch)
+        await communicate.save(output_path)
+        return output_path
         
-    # مدخلات إضافية ليتوافق مع طلب التطبيق
-    emotion_input = gr.Textbox(label="Emotion", value="محايد", visible=False)
-    symbol_input = gr.Checkbox(label="Is Symbol", value=True, visible=False)
-    
-    # المعاملات الجديدة المهمة
-    rate_input = gr.Textbox(label="Rate", value="+0%")
-    pitch_input = gr.Textbox(label="Pitch", value="+0Hz")
-    
-    output_audio = gr.Audio(label="Output", type="filepath")
-    
-    submit_btn = gr.Button("Generate")
-    
-    # ربط الدالة بالواجهة - تأكد من الترتيب الصحيح للمدخلات
-    submit_btn.click(
-        text_to_speech_edge, 
-        inputs=[text_input, voice_input, emotion_input, symbol_input, rate_input, pitch_input], 
-        outputs=output_audio
-    )
+    except Exception as e:
+        print(f"CRITICAL ERROR: {str(e)}")
+        # Return None or raise a Gradio Error to notify the frontend
+        raise gr.Error(f"Generation Failed: {str(e)}")
+
+# Define the Interface
+# The order of inputs MUST match the array sent from React:
+# [text, voice, emotion, is_symbol, rate, pitch]
+inputs = [
+    gr.Textbox(label="Text"),
+    gr.Dropdown(label="Voice", choices=list(VOICE_MAP.keys()) + list(VOICE_MAP.values())),
+    gr.Textbox(label="Emotion", value="neutral"), 
+    gr.Checkbox(label="Is Symbol", value=True),   
+    gr.Textbox(label="Rate", value="+0%"),
+    gr.Textbox(label="Pitch", value="+0Hz")
+]
+
+outputs = gr.Audio(label="Generated Audio", type="filepath")
+
+demo = gr.Interface(
+    fn=text_to_speech_edge,
+    inputs=inputs,
+    outputs=outputs,
+    title="Natiq Pro API",
+    allow_flagging="never"
+)
 
 if __name__ == "__main__":
     demo.queue().launch()