Update app.py

app.py

@@ -49,16 +49,15 @@ swiper_code = """
 st.components.v1.html(swiper_code, height=400)
 
 # Function to process speech-to-text
-def speech_to_text(audio_path):
-    with open(audio_path, "rb") as f:
-        files = {"file": f}
-        response = requests.post(HF_SPEECH_TO_TEXT_API, headers=HEADERS, files=files)
-        text = response.json().get("text", "Sorry, could not transcribe audio.")
+def speech_to_text(audio_bytes):
+    files = {"file": ("audio.wav", audio_bytes, "audio/wav")}
+    response = requests.post(HF_SPEECH_TO_TEXT_API, headers=HEADERS, files=files)
+    text = response.json().get("text", "Sorry, could not transcribe audio.")
     return text
 
 # Function to get chatbot response
 def chatbot_response(user_input):
-    payload = {"inputs": f"User: {user_input}\nAI:"}
+    payload = {"inputs": f"You are a friendly AI coach for English learners. Help them practice speaking naturally with supportive feedback.\nUser: {user_input}\nAI:"}
     response = requests.post(HF_CHATBOT_API, headers=HEADERS, json=payload)
     ai_reply = response.json().get("generated_text", "I'm here to help! Keep practicing.")
     return ai_reply
@@ -68,13 +67,32 @@ def text_to_speech(text):
     payload = {"inputs": text}
     response = requests.post(HF_TEXT_TO_SPEECH_API, headers=HEADERS, json=payload)
     return response.content
-
 # Streamlit UI
-st.title("🎙️ AI Speaking Coach - Talking Pals")
+st.title("🎙️ AI Speaking Pal")
+st.write("Press the button and chat with your pal!")
+
+# Audio Recording Component
+audio_input = st.audio_input("Let's talk!")
+
+if audio_input is not None:
+    # Read audio data
+    audio_bytes = audio_input.read()
+
+    # Process speech-to-text
+    user_text = speech_to_text(audio_bytes)
+    st.write(f"**You:** {user_text}")
+
+    # Get AI response
+    ai_reply = chatbot_response(user_text)
+    st.write(f"**AI:** {ai_reply}")
+
+    # Convert AI response to speech
+    speech_audio = text_to_speech(ai_reply)
+    st.audio(speech_audio, format="audio/wav")
 
 # Swiper Carousel Component
 avatar_list = list(avatars.keys())
-selected_index = st.slider("Pick your speaking coach:", 0, len(avatar_list) - 1, 0)
+selected_index = st.slider("Pick your pal:", 0, len(avatar_list) - 1, 0)
 selected_avatar = avatar_list[selected_index]
 avatar_info = avatars[selected_avatar]
 
@@ -103,7 +121,7 @@ if show_text:
 
 # Handle microphone input
 if st.session_state.conversation_active:
-    audio = audiorecorder("Click to Speak", "Stop Recording")
+    audio = audiorecorder("Click to Speak", "Stop recording")
     if len(audio) > 0:
         with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmpfile:
             tmpfile.write(audio)
@@ -114,4 +132,4 @@ if st.session_state.conversation_active:
             speech_audio = text_to_speech(ai_reply)
             st.audio(io.BytesIO(speech_audio), format="audio/wav")
 
-st.write("Have fun learning English with your AI speaking coach!")
+st.write("Have fun learning English with your AI pal!")