Update app.py

app.py

@@ -1,62 +1,70 @@
 import gradio as gr
 from transformers import pipeline
 
-# ——— Load Pipelines ———
+# ——— Load models ———
 chat_pipe = pipeline(
     "text-generation",
     model="Hulk810154/Kai",
     trust_remote_code=True
-)
+)  # Text generation 6
+
 asr_pipe = pipeline(
     "automatic-speech-recognition",
     model="openai/whisper-tiny"
-)
+)  # Whisper Tiny for STT 7
 
-# ——— Handlers ———
+# ——— Core handlers ———
 def chat_fn(message, history):
-    prompt = ""
-    for u, b in history:
-        prompt += f"User: {u}\nAI: {b}\n"
+    """Generate reply given message + history."""
+    prompt = "".join(f"User: {u}\nAI: {b}\n" for u, b in history)
     prompt += f"User: {message}\nAI:"
     out = chat_pipe(prompt, max_length=256, do_sample=True)
     reply = out[0]["generated_text"][len(prompt):].strip()
     history.append((message, reply))
     return history, history
 
-def speech_to_text(audio_path):
-    if not audio_path:
+def transcribe(audio):
+    """Turn recorded audio into text."""
+    if audio is None:
         return ""
-    return asr_pipe(audio_path)["text"]
+    # gr.Audio returns (array, sampling_rate) by default 8
+    if isinstance(audio, (tuple, list)) and len(audio) == 2:
+        arr, sr = audio
+        return asr_pipe({"array": arr, "sampling_rate": sr})["text"]
+    # Or accept a filepath
+    return asr_pipe(audio)["text"]
 
 def handle_upload(files):
-    return f"Received {len(files)} file(s): " + ", ".join([f.name for f in files])
+    """Report uploaded files/images."""
+    return f"Received {len(files)} file(s): " + ", ".join(f.name for f in files)
 
-# ——— UI ———
-with gr.Blocks() as demo:
-    gr.Markdown("# 🧠 Kai AGI Chat\n**Type, Speak or Upload — true multimodal AGI.**")
-    state = gr.State([])
+# ——— Build UI ———
+with gr.Blocks() as demo:  # Low-level Blocks API 9
+    gr.Markdown("# 🧠 Kai AGI Chat\n_Chat via text, voice, or upload — true multimodal AGI_")
+    state = gr.State([])  # Session memory
 
     with gr.Row():
         with gr.Column():
-            chatbot = gr.Chatbot(label="Conversation")
-            txt = gr.Textbox(placeholder="Type here…", label="Text Input")
-            mic = gr.Audio(source="microphone", type="filepath", label="🎤 Voice Input")
-            send = gr.Button("Send")
+            chatbot = gr.Chatbot(type="messages", label="Conversation")  # Future-proof format 10
+            txt_input = gr.Textbox(placeholder="Type here…", label="Text Input")
+            mic_input = gr.Audio(label="🎤 Voice Input")  # Browser mic/upload 11
+            send_btn = gr.Button("Send")
         with gr.Column():
             uploader = gr.File(file_count="multiple", label="📎 Upload Files/Images")
             upload_status = gr.Textbox(label="Upload Status")
 
-    # Wire it up
-    send.click(
-        lambda txt, mic, history: ( speech_to_text(mic) or txt, history ),
-        inputs=[txt, mic, state],
-        outputs=[txt, state]
+    # Text or voice → chat
+    send_btn.click(
+        lambda txt, mic, hist: (transcribe(mic) or txt, hist),
+        inputs=[txt_input, mic_input, state],
+        outputs=[txt_input, state]
     ).then(
         chat_fn,
-        inputs=[txt, state],
+        inputs=[txt_input, state],
         outputs=[chatbot, state]
     )
 
+    # Handle uploads
     uploader.upload(handle_upload, inputs=uploader, outputs=upload_status)
 
 demo.launch()