Eason918's picture
Update app.py
239d642 verified
raw
history blame
2.38 kB
import streamlit as st
from PIL import Image
from transformers import pipeline
from gtts import gTTS
import os
# ๅŠ ่ผ‰ Hugging Face ๆจกๅž‹
image_to_text_model = pipeline("image-to-text", model="Salesforce/blip-image-captioning-base")
story_generator = pipeline("text-generation", model="facebook/opt-1.3b")
# ๅœ–็‰‡ โ†’ ๆ–‡ๅญ—๏ผˆ็”Ÿๆˆๆ่ฟฐ๏ผ‰
def img2text(image_path):
text = image_to_text_model(image_path)[0]["generated_text"]
return text
# ๆ–‡ๅญ— โ†’ ๆ•…ไบ‹๏ผˆ็”ŸๆˆๅฎŒๆ•ดๆ•…ไบ‹๏ผ‰
def text2story(text):
prompt = f"Write a fun and magical children's story based on this idea: {text}.\n\nOnce upon a time..."
story = story_generator(prompt, max_length=250, do_sample=True, temperature=0.8, top_p=0.9, repetition_penalty=1.2, truncation=True)[0]['generated_text']
return story
# ๆ•…ไบ‹ โ†’ ่ชž้Ÿณ๏ผˆTTS๏ผ‰
def text2audio_gtts(story_text, filename="story.mp3"):
# ้ฟๅ…ๆ–‡ไปถๅ†ฒ็ช
if os.path.exists(filename):
os.remove(filename)
# ้™ๅˆถ TTS ๆ–‡ๆœฌ้•ทๅบฆ
max_chars = 500 # gTTS ๅฏ่ƒฝไธๆ”ฏๆŒ้Ž้•ทๆ–‡ๆœฌ
story_text = story_text[:max_chars]
# ็”Ÿๆˆ่ฏญ้Ÿณ
tts = gTTS(text=story_text, lang="en")
tts.save(filename)
return filename
# Streamlit Web UI
st.set_page_config(page_title="AI Storyteller", page_icon="๐Ÿ“–")
st.header("๐Ÿ“– AI Storyteller: Turn Your Image into a Story with Audio")
uploaded_file = st.file_uploader("Upload an Image...", type=["jpg", "png"])
if uploaded_file:
# ไฟๅญ˜ๅœ–็‰‡ๅˆฐๆœฌๅœฐ
image_path = "uploaded_image.jpg"
with open(image_path, "wb") as f:
f.write(uploaded_file.getbuffer())
# ่ฎ€ๅ–ไธฆ้กฏ็คบๅœ–็‰‡
image = Image.open(image_path)
st.image(image, caption="Uploaded Image", use_column_width=True)
# ็”Ÿๆˆๅœ–็‰‡ๆ่ฟฐ
st.text("๐Ÿ” Generating image caption...")
caption = img2text(image_path) # ้€™่ฃๆ”นๆˆๆ–‡ไปถ่ทฏๅพ‘
st.write("**Image Description:**", caption)
# ็”Ÿๆˆๆ•…ไบ‹
st.text("๐Ÿ“ Generating story...")
story = text2story(caption)
st.write("**Generated Story:**")
st.write(story)
# ็”Ÿๆˆ่ชž้Ÿณ
st.text("๐Ÿ”Š Generating audio...")
audio_file = text2audio_gtts(story)
# ๆ’ญๆ”พ้Ÿณ้ ป
st.audio(audio_file, format="audio/mp3")
# ไธ‹่ผ‰ๆŒ‰้’ฎ
with open(audio_file, "rb") as file:
st.download_button("๐Ÿ“ฅ Download Audio", file, file_name="story.mp3")