Update app.py

app.py

@@ -2,22 +2,38 @@ import os
 import asyncio
 import wave
 import gradio as gr
+import tempfile
+from groq import Groq
 from google import genai
 from google.genai import types
-import tempfile
 
 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}."
+PATH = os.environ.get("PATH")
 MODEL_ID = os.environ.get("MODEL_VERSION")
 UK_SERVER_API = os.environ.get("UK_SERVER_API")
+GROQ_API_KEY = os.environ.get("GROQ_API_KEY")
+
+client_gemini = genai.Client(api_key=UK_SERVER_API)
+client_groq = Groq(api_key=GROQ_API_KEY)
 
-client = genai.Client(api_key=UK_SERVER_API)
+async def nilla_engine(audio_path):
+    if audio_path is None:
+        return None, "No audio provided"
 
-async def nilla_voice_engine(text):
-    temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
-    file_path = temp_file.name
+    with open(audio_path, "rb") as file:
+        transcription = client_groq.audio.transcriptions.create(
+            file=(audio_path, file.read()),
+            model="whisper-large-v3",
+            temperature=0,
+            response_format="verbose_json",
+        )
+    
+    user_text = transcription.text
+    
+    output_temp = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
+    output_path = output_temp.name
     
     config = {
         "response_modalities": ["AUDIO"],
@@ -25,13 +41,13 @@ async def nilla_voice_engine(text):
     }
     
     try:
-        async with client.aio.live.connect(model=MODEL_ID, config=config) as session:
+        async with client_gemini.aio.live.connect(model=MODEL_ID, config=config) as session:
             await session.send_client_content(
-                turns={"role": "user", "parts": [{"text": text}]}, 
+                turns={"role": "user", "parts": [{"text": user_text}]}, 
                 turn_complete=True
             )
 
-            with wave.open(file_path, "wb") as wav:
+            with wave.open(output_path, "wb") as wav:
                 wav.setnchannels(1)
                 wav.setsampwidth(2)
                 wav.setframerate(24000)
@@ -41,26 +57,32 @@ async def nilla_voice_engine(text):
                     if response.data is not None:
                         wav.writeframes(response.data)
         
-        return file_path
-    except Exception as e:
-        return None
+        return output_path, user_text
+    except Exception:
+        return None, user_text
 
-def run_api(text):
-    if not text:
-        return None
+def run_interface(audio_file):
     loop = asyncio.new_event_loop()
     asyncio.set_event_loop(loop)
-    return loop.run_until_complete(nilla_voice_engine(text))
+    return loop.run_until_complete(nilla_engine(audio_file))
 
 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")
+    with gr.Row():
+        input_audio = gr.Audio(label="Speak to Nilla", type="filepath")
+    
+    with gr.Row():
+        output_text = gr.Textbox(label="Transcribed Text (Groq)")
+        output_audio = gr.Audio(label="Nilla's Response")
     
-    submit_btn.click(fn=run_api, inputs=input_text, outputs=output_audio)
+    submit_btn = gr.Button("Process")
+    submit_btn.click(
+        fn=run_interface, 
+        inputs=input_audio, 
+        outputs=[output_audio, output_text]
+    )
 
 if __name__ == "__main__":
     demo.launch()