app.py

import streamlit as st
from streamlit.components.v1 import html
import requests
import os
import base64
import tempfile
import io
from audiorecorder import audiorecorder
import streamlit.components.v1 as components
from PIL import Image

# Hugging Face API Keys
HF_API_KEY = os.getenv("HF_API_KEY")
HEADERS = {"Authorization": f"Bearer {HF_API_KEY}"}

# Hugging Face API Endpoints
HF_SPEECH_TO_TEXT_API = "https://api-inference.huggingface.co/models/openai/whisper-small"
HF_CHATBOT_API = "https://api-inference.huggingface.co/models/mistralai/Mistral-7B-Instruct"
HF_TEXT_TO_SPEECH_API = "https://api-inference.huggingface.co/models/facebook/mms-tts-eng"

# Avatars and their personalities
avatars = {
    "rat": {"desc": "Clever and resourceful. Loves puzzles and word games.", "voice": "Hello! Let's solve some fun puzzles together!"},
    "ox": {"desc": "Steady and dependable. Encourages patience in learning.", "voice": "Consistency is key! Let's learn step by step."},
    "tiger": {"desc": "Bold and adventurous. Loves dynamic conversations.", "voice": "Let's jump into an exciting conversation!"},
    "rabbit": {"desc": "Gentle and kind. A patient and encouraging teacher.", "voice": "Take your time. Learning should be fun!"},
    "dragon": {"desc": "Wise and powerful. Inspires confidence in learning.", "voice": "Believe in yourself! You can do this!"},
    "snake": {"desc": "Calm and analytical. Gives insightful explanations.", "voice": "Let's explore the beauty of language together."},
    "horse": {"desc": "Energetic and friendly. Encourages active learning.", "voice": "Let's have fun while we learn!"},
    "goat": {"desc": "Gentle and artistic. Uses creativity to teach.", "voice": "Imagine and express yourself freely!"},
    "monkey": {"desc": "Playful and clever. Loves challenges and fun exercises.", "voice": "I love games! Let's play with words!"},
    "rooster": {"desc": "Confident and expressive. Loves discussions.", "voice": "Speak up! Let's share ideas together."},
    "dog": {"desc": "Loyal and friendly. Encourages persistence.", "voice": "I'll always be here to help you practice!"},
    "pig": {"desc": "Warm and cheerful. Creates a positive learning atmosphere.", "voice": "Learning is fun! Let's enjoy this together!"},
}

# Load the Swiper component
swiper_code = """
<script type="module">
    import React from "react";
    import ReactDOM from "react-dom";
    import SwiperComponent from "./Swipercomponent.js";
    function App() {
        return <SwiperComponent />;
    }
    ReactDOM.render(<App />, document.getElementById("root"));
</script>
"""
st.components.v1.html(swiper_code, height=400)

# Function to process speech-to-text
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"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

# Function for text-to-speech
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 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 pal:", 0, len(avatar_list) - 1, 0)
selected_avatar = avatar_list[selected_index]
avatar_info = avatars[selected_avatar]

# Display Avatar Image
avatar_img = Image.open(f"images/{selected_avatar}.png")
st.image(avatar_img, caption=f"{selected_avatar.capitalize()} - {avatar_info['desc']}", use_column_width=True)

# Play avatar's sample voice
if st.button("Hear Sample"): 
    audio_bytes = text_to_speech(avatar_info["voice"])
    st.audio(io.BytesIO(audio_bytes), format="audio/wav")

# Start/Stop Conversation Button
if "conversation_active" not in st.session_state:
    st.session_state.conversation_active = False

def toggle_conversation():
    st.session_state.conversation_active = not st.session_state.conversation_active

st.button("🎤 Start/Stop Conversation", on_click=toggle_conversation)

# Conversation Text Box
show_text = st.checkbox("Show conversation text")
if show_text:
    conversation_box = st.text_area("Conversation:", height=200)

# Handle microphone input
if st.session_state.conversation_active:
    audio = audiorecorder("Click to Speak", "Stop recording")
    if len(audio) > 0:
        with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmpfile:
            tmpfile.write(audio)
            user_text = speech_to_text(tmpfile.name)
            st.write(f"**You:** {user_text}")
            ai_reply = chatbot_response(user_text)
            st.write(f"**{selected_avatar.capitalize()}:** {ai_reply}")
            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 pal!")