Update app.py

app.py

@@ -1,57 +1,88 @@
-
 import streamlit as st
-from PIL import Image
-import torch
-import torchvision.transforms as transforms
-from animegan2.model import Generator
+import tempfile
 import os
-import urllib.request
-
-st.set_page_config(page_title="Ghibli Style Image Converter 🎨")
-
-MODEL_PATH = "weights/paprika.pt"
-MODEL_URL = "https://huggingface.co/akhaliq/AnimeGANv2/resolve/main/paprika.pt"
-
-@st.cache_resource
-def load_model():
-    if not os.path.exists(MODEL_PATH):
-        os.makedirs("weights", exist_ok=True)
-        with st.spinner("Downloading model weights..."):
-            urllib.request.urlretrieve(MODEL_URL, MODEL_PATH)
-            st.success("Model downloaded successfully!")
-    model = Generator()
-    model.load_state_dict(torch.load(MODEL_PATH, map_location="cpu"))
-    model.eval()
-    return model
-
-def preprocess(image):
-    image = image.convert("RGB")
-    transform = transforms.Compose([
-        transforms.Resize((256, 256)),
-        transforms.ToTensor()
-    ])
-    return transform(image).unsqueeze(0)
-
-def postprocess(tensor):
-    tensor = tensor.squeeze().detach().cpu().clamp(0, 1)
-    image = transforms.ToPILImage()(tensor)
-    return image
-
-st.title("🎨 Convert to Ghibli Style!")
-st.markdown("Upload an image to transform it into a beautiful Ghibli-style artwork using AnimeGAN2!")
-
-uploaded_file = st.file_uploader("Upload your image", type=["jpg", "jpeg", "png"])
-
-if uploaded_file:
-    input_image = Image.open(uploaded_file)
-    st.image(input_image, caption="Original Image", use_column_width=True)
-
-    with st.spinner("Converting... please wait!"):
-        model = load_model()
-        input_tensor = preprocess(input_image)
-        with torch.no_grad():
-            output_tensor = model(input_tensor)
-        output_image = postprocess(output_tensor)
-
-    st.image(output_image, caption="Ghibli Style Image", use_column_width=True)
-    st.success("Done!")
+import uuid
+import torch
+import json
+from datetime import datetime
+from transformers import pipeline
+import coqui_tts
+from TTS.api import TTS
+import whisper
+import requests
+
+# Initialize Whisper model (tiny)
+whisper_model = whisper.load_model("tiny")
+
+# Load TTS model (Coqui TTS)
+tts = TTS(model_name="tts_models/en/vctk/vits")
+
+# Conversation history
+if "history" not in st.session_state:
+    st.session_state.history = []
+
+st.set_page_config(page_title="Voice Chat App", layout="centered")
+st.title("🗣️ Voice-Based Conversational App")
+
+st.sidebar.header("🎛️ Settings")
+language = st.sidebar.selectbox("Select Language", ["en", "es", "fr", "de", "it"])
+emotion = st.sidebar.selectbox("Select Emotion", ["neutral", "happy", "sad", "angry"])
+voice_avatar = st.sidebar.selectbox("Select Voice Avatar", ["female", "male"])
+
+st.markdown("Speak or upload a voice file to start chatting with the AI")
+
+# Voice input
+voice_input = st.file_uploader("Upload a voice file", type=["wav", "mp3", "m4a"])
+
+if voice_input:
+    # Save temporary file
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_file:
+        tmp_file.write(voice_input.read())
+        tmp_path = tmp_file.name
+
+    # Transcribe using Whisper
+    st.info("Transcribing...")
+    result = whisper_model.transcribe(tmp_path, language=language)
+    user_text = result["text"]
+    st.success(f"You said: {user_text}")
+
+    # Chat with Groq API (LLaMA3 or Mixtral)
+    st.info("Generating response...")
+    GROQ_API_KEY = os.getenv("GROQ_API_KEY")
+    headers = {
+        "Authorization": f"Bearer {GROQ_API_KEY}",
+        "Content-Type": "application/json"
+    }
+    data = {
+        "model": "mixtral-8x7b-32768",
+        "messages": [{"role": "user", "content": user_text}],
+        "temperature": 0.7
+    }
+    response = requests.post("https://api.groq.com/openai/v1/chat/completions", headers=headers, data=json.dumps(data))
+    reply_text = response.json()["choices"][0]["message"]["content"]
+    st.success(f"AI: {reply_text}")
+
+    # Save history
+    st.session_state.history.append({"user": user_text, "bot": reply_text})
+
+    # Convert to speech
+    st.info("Generating voice reply...")
+    output_path = f"output_{uuid.uuid4().hex}.wav"
+    tts.tts_to_file(text=reply_text, file_path=output_path, speaker=0 if voice_avatar == "female" else 1)
+
+    # Play response
+    audio_file = open(output_path, "rb")
+    audio_bytes = audio_file.read()
+    st.audio(audio_bytes, format="audio/wav")
+
+    # Clean up
+    os.remove(tmp_path)
+    os.remove(output_path)
+
+# Show conversation history
+if st.session_state.history:
+    st.subheader("📜 Conversation History")
+    for i, item in enumerate(st.session_state.history):
+        st.markdown(f"**You:** {item['user']}")
+        st.markdown(f"**AI:** {item['bot']}")
+        st.markdown("---")