Update app.py

app.py

@@ -1,48 +1,103 @@
+# Imports
 import os
-import requests
 import streamlit as st
+import requests
+from transformers import pipeline
+import openai
+
+# Suppressing all warnings
+import warnings
+warnings.filterwarnings("ignore")
+
+# Image-to-text
+def img2txt(url):
+    print("Initializing captioning model...")
+    captioning_model = pipeline("image-to-text", model="Salesforce/blip-image-captioning-base")
+    
+    print("Generating text from the image...")
+    text = captioning_model(url, max_new_tokens=20)[0]["generated_text"]
+    
+    print(text)
+    return text
+
+# Text-to-story
+def txt2story(img_text):
+    print("Initializing client...")
+    client = openai.OpenAI(
+        api_key=os.environ["TOGETHER_API_KEY"], 
+        base_url='https://api.together.xyz',
+    )
+    
+    messages = [
+        {"role": "system", "content": '''As an experienced short story writer, write story title and then create a meaningful story influenced by provided words. 
+        Ensure stories conclude positively within 100 words. Remember the story must end within 100 words''', "temperature": 1.8},
+        {"role": "user", "content": f"Here is input set of words: {img_text}", "temperature": 1.5},
+    ]
+    
+    print("Story...")
+    chat_completion = client.chat.completions.create(
+        messages=messages,
+        model="togethercomputer/llama-2-70b-chat")
+    
+    print(chat_completion.choices[0].message.content)
+    return chat_completion.choices[0].message.content
+
+
+
+# Text-to-speech
+def txt2speech(text):
+    print("Initializing text-to-speech conversion...")
+    API_URL = "https://api-inference.huggingface.co/models/espnet/kan-bayashi_ljspeech_vits"
+    headers = {"Authorization": f"Bearer {os.environ['HUGGINGFACEHUB_API_TOKEN']}"}
+    payloads = {'inputs': text}
+
+    response = requests.post(API_URL, headers=headers, json=payloads)
+    
+    with open('audio_story.mp3', 'wb') as file:
+        file.write(response.content)
+
+        
+# Streamlit web app main function
+def main():
+    st.set_page_config(page_title="🎨 Image-to-Audio Story 🎧", page_icon="🖼️")
+    st.title("Turn the Image into Audio Story")
+
+# Allows users to upload an image file
+    uploaded_file = st.file_uploader("# 📷 Upload an image...", type=["jpg", "jpeg", "png"])
+
+    if uploaded_file is not None:
+# Reads and saves uploaded image file
+        bytes_data = uploaded_file.read()
+        with open("uploaded_image.jpg", "wb") as file:
+            file.write(bytes_data)
+
+        st.image(uploaded_file, caption='🖼️ Uploaded Image', use_column_width=True)
+
+ # Initiates AI processing and story generation
+        with st.spinner("## 🤖 AI is at Work! "):
+            scenario = img2txt("uploaded_image.jpg")  # Extracts text from the image
+            story = txt2story(scenario)  # Generates a story based on the image text
+            txt2speech(story)  # Converts the story to audio
+
+            st.markdown("---")
+            st.markdown("## 📜 Image Caption")
+            st.write(scenario)
+
+            st.markdown("---")
+            st.markdown("## 📖 Story")
+            st.write(story)
+
+            st.markdown("---")
+            st.markdown("## 🎧 Audio Story")
+            st.audio("audio_story.mp3")
+             
+if __name__ == '__main__':
+    main()
 
-# API details
-API_URL = "https://api-inference.huggingface.co/models/openai-community/gpt2-xl"
-API_TOKEN = os.environ.get('HUGGINGFACEHUB_API_TOKEN')
-HEADERS = {"Authorization": f"Bearer {API_TOKEN}"}
-
-# Streamlit UI
-st.title("GPT-2 Movie Sentiment Analysis")
-
-# Input text for sentiment analysis
-input_text = st.text_area("Enter movie review:", "")
-
-# Choose analysis type
-analysis_type = st.radio("Select analysis type:", ["Zero-shot", "One-shot", "Few-shot"])
-
-if st.button("Analyze Sentiment"):
-    # Prepare payload for API request
-    if analysis_type == "Zero-shot":
-        payload = {"inputs": f"Label the text as either 'positive', 'negative', or 'mixed' related to a movie:\n\n{input_text}"}
-    elif analysis_type == "One-shot":
-        prompt = "Label the sentence as either 'positive', 'negative', or 'mixed' related to a movie:\n\n" \
-                 "Sentence: This movie exceeded my expectations.\nLabel: positive"
-        payload = {"inputs": f"{prompt} {input_text}"}
-    elif analysis_type == "Few-shot":
-        examples = [
-            "Sentence: The cinematography in this movie is outstanding.\nLabel: positive",
-            "Sentence: I didn't enjoy the plot twists in the movie.\nLabel: negative",
-            "Sentence: The acting was great, but the pacing felt off.\nLabel: mixed",
-            "Sentence: This movie didn't live up to the hype.\nLabel: negative",
-        ]
-        prompt = "Label the sentences as either 'positive', 'negative', or 'mixed' related to a movie:\n\n" + "\n".join(examples)
-        payload = {"inputs": f"{prompt}\n\n{input_text}"}
-
-    # Make API request
-    response = requests.post(API_URL, headers=HEADERS, json=payload)
-
-    # Print entire response for debugging
-    st.write("API Response:", response.json())
-
-    # Display results
-    if response.status_code == 200:
-        result = response.json()[0]  # Assuming the sentiment is in the first item of the list
-        st.write("Sentiment:", result.get('generated_text', 'N/A'))
-    else:
-        st.write("Error:", response.status_code, response.text)
+# Credits
+st.markdown("### Credits")
+st.caption('''
+            Made with ❤️ by @Aditya-Neural-Net-Ninja\n 
+            Utilizes Image-to-text, Text Generation, Text-to-speech Transformer Models\n
+            Gratitude to Streamlit, 🤗 Spaces for Deployment & Hosting
+            ''')