Update src/streamlit_app.py

src/streamlit_app.py

@@ -4,7 +4,8 @@ import torch
 from langdetect import detect
 from PIL import Image
 import numpy as np
-import pyttsx3
+from gtts import gTTS
+import tempfile
 
 # ----------------------------
 # Load Model
@@ -24,13 +25,14 @@ def load_model():
 tokenizer, model = load_model()
 
 # ----------------------------
-# TTS (Offline)
+# Text-to-Speech (HF SAFE)
 # ----------------------------
-engine = pyttsx3.init()
+def speak(text, lang="en"):
+    tts = gTTS(text=text, lang=lang)
 
-def speak(text):
-    engine.say(text)
-    engine.runAndWait()
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as fp:
+        tts.save(fp.name)
+        return fp.name
 
 # ----------------------------
 # Language Detection
@@ -46,7 +48,7 @@ def detect_lang(text):
 # ----------------------------
 def build_prompt(user_input, lang):
     if lang == "fr":
-        system = "Tu es un tuteur de mathématiques pour enfants. Réponds simplement avec des exemples."
+        system = "Tu es un tuteur de mathématiques pour enfants. Explique simplement avec des exemples."
     elif lang == "sw":
         system = "Wewe ni mwalimu wa hesabu kwa watoto. Eleza kwa urahisi."
     else:
@@ -68,7 +70,7 @@ def generate(prompt):
     return tokenizer.decode(output[0], skip_special_tokens=True)
 
 # ----------------------------
-# Visual Counting (Lightweight)
+# Visual Counting
 # ----------------------------
 def count_objects(image):
     img = np.array(image.convert("L"))
@@ -77,21 +79,22 @@ def count_objects(image):
     return max(1, count)
 
 # ----------------------------
-# UI DASHBOARD
+# UI CONFIG
 # ----------------------------
 st.set_page_config(layout="wide")
-st.title("🧠📊 AI Math Tutor Dashboard")
+st.title("🧠📊 AI Math Tutor (Multimodal)")
 
 col1, col2 = st.columns(2)
 
 # ----------------------------
-# LEFT PANEL — INTERACTION
+# LEFT PANEL — STUDENT
 # ----------------------------
 with col1:
     st.header("👧 Student Interaction")
 
     mode = st.radio("Choose Mode", ["Text", "Image (Count)", "Voice (Simulated)"])
 
+    # ---------------- TEXT MODE ----------------
     if mode == "Text":
         user_input = st.text_input("Ask a math question:")
 
@@ -99,17 +102,17 @@ with col1:
             lang = detect_lang(user_input)
             prompt = build_prompt(user_input, lang)
 
-            response = generate(prompt)
+            with st.spinner("Thinking..."):
+                response = generate(prompt)
 
             st.write("### 📘 Answer")
             st.write(response)
 
             if st.button("🔊 Speak Answer"):
-                speak(response)
+                audio_file = speak(response, lang)
+                st.audio(audio_file)
 
-    # ----------------------------
-    # IMAGE MODE (Visual Learning)
-    # ----------------------------
+    # ---------------- IMAGE MODE ----------------
     elif mode == "Image (Count)":
         uploaded = st.file_uploader("Upload image with objects", type=["png", "jpg"])
 
@@ -119,52 +122,50 @@ with col1:
 
             count = count_objects(image)
 
-            st.write(f"### 🧮 I see about: {count} objects")
-
-            explanation = f"There are about {count} objects. Let's count together!"
-            st.write(explanation)
+            explanation = f"I see about {count} objects. Let's count together!"
+            st.write(f"### 🧮 {explanation}")
 
             if st.button("🔊 Speak"):
-                speak(explanation)
+                audio_file = speak(explanation)
+                st.audio(audio_file)
 
-    # ----------------------------
-    # VOICE MODE (SIMULATED)
-    # ----------------------------
+    # ---------------- VOICE MODE ----------------
     elif mode == "Voice (Simulated)":
-        st.write("🎤 Voice input simulation (type what child says)")
+        st.write("🎤 Simulate child's speech (type it)")
 
         voice_input = st.text_input("Child says:")
 
         if voice_input:
             lang = detect_lang(voice_input)
-
             prompt = build_prompt(voice_input, lang)
+
             response = generate(prompt)
 
             st.write("### 🎧 Tutor Response")
             st.write(response)
 
             if st.button("🔊 Speak Response"):
-                speak(response)
+                audio_file = speak(response, lang)
+                st.audio(audio_file)
 
 # ----------------------------
-# RIGHT PANEL — PROGRESS
+# RIGHT PANEL — DASHBOARD
 # ----------------------------
 with col2:
-    st.header("📈 Learning Progress")
+    st.header("📈 Learning Dashboard")
 
-    # Fake metrics (replace with SQLite later)
+    # Demo metrics (replace with SQLite later)
     st.metric("Questions Answered", 12)
     st.metric("Accuracy", "75%")
     st.metric("Level", "Beginner → Improving")
 
-    st.subheader("📊 Skill Breakdown")
+    st.subheader("📊 Skill Progress")
     st.progress(0.7)
 
-    st.subheader("🌍 Language Detected")
-    st.write("Auto-detected per input")
+    st.subheader("🌍 Language Handling")
+    st.write("Auto-detect + multilingual response")
 
-    st.subheader("⚡ System Info")
-    st.write("CPU Mode ✔")
-    st.write("Offline-ready ✔")
-    st.write("Multilingual ✔")
+    st.subheader("⚡ System Status")
+    st.success("CPU Mode")
+    st.success("Multimodal Enabled")
+    st.success("Deployable on Hugging Face")