Update app.py

app.py

@@ -1,29 +1,66 @@
+import os
+import asyncio
+import wave
 import gradio as gr
-from faster_whisper import WhisperModel
+from google import genai
+from google.genai import types
+import tempfile
 
-# انتخاب مدل: 'base' یا 'small' برای CPU عالی هستند
-# 'int8' باعث می‌شود مدل روی CPU بسیار سریع و کم‌حجم اجرا شود
-model_size = "small"
-model = WhisperModel(model_size, device="cpu", compute_type="int8")
+BOT_NAME = "Nilla"
+MOTOR_NAME = "Nilla-2026 GPT motor"
+PROVIDER = "HumanV lab"
+SYSTEM_INSTRUCTION = f"Your name is {BOT_NAME} with {MOTOR_NAME} and providing by {PROVIDER}."
+MODEL_ID = os.environ.get("MODEL_VERSION")
+UK_SERVER_API = os.environ.get("UK_SERVER_API")
 
-def transcribe_audio(audio):
-    # اجرای تبدیل صدا به متن
-    # beam_size=5 دقت را بالا می‌برد
-    segments, info = model.transcribe(audio, beam_size=5)
+client = genai.Client(api_key=UK_SERVER_API)
+
+async def nilla_voice_engine(text):
+    temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
+    file_path = temp_file.name
+    
+    config = {
+        "response_modalities": ["AUDIO"],
+        "system_instruction": SYSTEM_INSTRUCTION
+    }
     
-    full_text = ""
-    for segment in segments:
-        full_text += segment.text + " "
+    try:
+        async with client.aio.live.connect(model=MODEL_ID, config=config) as session:
+            await session.send_client_content(
+                turns={"role": "user", "parts": [{"text": text}]}, 
+                turn_complete=True
+            )
+
+            with wave.open(file_path, "wb") as wav:
+                wav.setnchannels(1)
+                wav.setsampwidth(2)
+                wav.setframerate(24000)
+
+                turn = session.receive()
+                async for response in turn:
+                    if response.data is not None:
+                        wav.writeframes(response.data)
         
-    return full_text
-
-# طراحی رابط کاربری
-demo = gr.Interface(
-    fn=transcribe_audio,
-    inputs=gr.Audio(type="filepath"),
-    outputs="text",
-    title="تبدیل رایگان صدا به متن (Whisper CPU)",
-    description="فایل صوتی خود را آپلود کنید تا با دقت بالا به متن تبدیل شود. پشتیبانی از تمامی زبان‌ها از جمله فارسی."
-)
-
-demo.launch()
+        return file_path
+    except Exception as e:
+        return None
+
+def run_api(text):
+    if not text:
+        return None
+    loop = asyncio.new_event_loop()
+    asyncio.set_event_loop(loop)
+    return loop.run_until_complete(nilla_voice_engine(text))
+
+with gr.Blocks(title=BOT_NAME) as demo:
+    gr.Markdown(f"# {BOT_NAME}")
+    gr.Markdown(f"{MOTOR_NAME} | {PROVIDER}")
+    
+    input_text = gr.Textbox(label="Input", placeholder="...")
+    output_audio = gr.Audio(label="Response", type="filepath")
+    submit_btn = gr.Button("Generate")
+    
+    submit_btn.click(fn=run_api, inputs=input_text, outputs=output_audio)
+
+if __name__ == "__main__":
+    demo.launch()