Update app.py

app.py

@@ -9,16 +9,17 @@ from transformers import pipeline
 # =============================
 # 1. Hugging Face Authentication
 # =============================
-HF_TOKEN = os.getenv("HF_TOKEN")  # Make sure to set: export HF_TOKEN="your_token_here"
+HF_TOKEN = os.getenv("HF_TOKEN")  # export HF_TOKEN="your_token_here"
 if HF_TOKEN is None:
     raise ValueError("⚠️ Please set your HF_TOKEN as an environment variable.")
 
 # =============================
-# 2. Load embedding + QA model
+# 2. Load Models
 # =============================
+# Embedding + QA
 embedding_model = SentenceTransformer(
     "sentence-transformers/all-MiniLM-L6-v2",
-    use_auth_token=HF_TOKEN 
+    use_auth_token=HF_TOKEN
 )
 qa_model = pipeline(
     "text-generation",
@@ -27,6 +28,20 @@ qa_model = pipeline(
     device_map="auto"
 )
 
+# Speech-to-Text (Whisper small, lightweight)
+stt_model = pipeline(
+    "automatic-speech-recognition",
+    model="openai/whisper-small",
+    token=HF_TOKEN
+)
+
+# Text-to-Speech (VITS)
+tts_model = pipeline(
+    "text-to-speech",
+    model="espnet/kan-bayashi_ljspeech_vits",
+    token=HF_TOKEN
+)
+
 # =============================
 # 3. Helper: extract text from files
 # =============================
@@ -95,13 +110,31 @@ def answer_query(query):
     return response
 
 # =============================
-# 8. Gradio UI (Visually Appealing)
+# 8. Voice-enabled Query
+# =============================
+def voice_query(audio):
+    if audio is None:
+        return "⚠️ Please record your question.", None
+
+    # Step 1: Speech-to-Text
+    stt_result = stt_model(audio)
+    text_query = stt_result["text"]
+
+    # Step 2: Get Answer from RAG
+    answer = answer_query(text_query)
+
+    # Step 3: Text-to-Speech
+    tts_result = tts_model(answer)
+    return answer, (tts_result["audio"], tts_result["sampling_rate"])
+
+# =============================
+# 9. Gradio UI (Visually Appealing)
 # =============================
 with gr.Blocks(theme=gr.themes.Soft(primary_hue="indigo", secondary_hue="cyan")) as demo:
     gr.Markdown("""
-    # 📚 HyDE RAG Chatbot
-    Talk with your documents using **Hypothetical Document Embeddings (HyDE)**.
-    Upload a PDF/DOCX/TXT and start asking questions!
+    # 📚 HyDE RAG Chatbot + 🎤 Voice Assistant
+    Talk with your documents using **Hypothetical Document Embeddings (HyDE)**.  
+    Upload a PDF/DOCX/TXT and start asking questions by **typing or speaking**!
     """)
 
     with gr.Row():
@@ -111,12 +144,18 @@ with gr.Blocks(theme=gr.themes.Soft(primary_hue="indigo", secondary_hue="cyan"))
             status = gr.Textbox(label="Status", interactive=False)
 
         with gr.Column(scale=2):
-            query = gr.Textbox(label="❓ Ask a Question", placeholder="Type your question here...")
+            query = gr.Textbox(label="❓ Ask a Question (Text)", placeholder="Type your question here...")
             ask_btn = gr.Button("🚀 Get Answer", variant="primary")
             answer = gr.Textbox(label="💡 Answer", lines=6)
 
+            gr.Markdown("### 🎤 Or Ask by Voice")
+            mic_input = gr.Audio(sources=["microphone"], type="filepath", label="Speak your question")
+            voice_answer = gr.Textbox(label="💡 Answer (from voice)", lines=6)
+            voice_output = gr.Audio(label="🔊 Bot Voice Reply")
+
+    # Events
     upload_btn.click(upload_file, inputs=file_input, outputs=status)
     ask_btn.click(answer_query, inputs=query, outputs=answer)
+    mic_input.change(voice_query, inputs=mic_input, outputs=[voice_answer, voice_output])
 
 demo.launch()
-