app.py

from transformers import BlipProcessor, BlipForConditionalGeneration
from PIL import Image
# Load model directly
from transformers import AutoProcessor, AutoModelForImageTextToText

def generate_caption(image_path):
    image = Image.open(image_path)
    processor = AutoProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
    model = AutoModelForImageTextToText.from_pretrained("Salesforce/blip-image-captioning-base")
    inputs = processor(image, return_tensors="pt")
    output = model.generate(**inputs)
    caption = processor.decode(output[0], skip_special_tokens=True)
    return caption
from transformers import pipeline

def generate_story(caption):
    # 使用文本生成 pipeline
    generator = pipeline("text-generation", model="gpt2")
    prompt = f"由以下图片得到的描述: '{caption}'，请根据这个描述生成一个完整的童话故事，故事至少100个单词。"
    result = generator(prompt, max_length=300, num_return_sequences=1)
    story = result[0]['generated_text']
    # 添加字数判断，必要时进行调整或循环生成直到满足条件
    if len(story.split()) < 100:
        # 可以进行递归调用或其他逻辑扩充文本
        story += " " + generate_story(caption)
    return story
from gtts import gTTS

def text_to_speech(text, output_file="output.mp3"):
    tts = gTTS(text=text, lang='en')  # 注意可根据需要选择语言或使用中文
    tts.save(output_file)
    return output_file
import streamlit as st
from PIL import Image

def main():
    st.title("儿童故事生成应用")
    st.write("上传一张图片，我们将根据图片生成有趣的故事，并转换成语音播放给你听！")
    
    uploaded_file = st.file_uploader("选择一张图片", type=["png", "jpg", "jpeg"])
    if uploaded_file is not None:
        image = Image.open(uploaded_file)
        st.image(image, caption="上传的图片", use_container_width=True)
        
        # 调用图像描述函数
        caption = generate_caption(uploaded_file)
        st.write("图片描述:", caption)
        
        # 生成故事
        story = generate_story(caption)
        st.write("生成的故事:", story)
        
        # 文字转语音
        audio_file = text_to_speech(story)
        st.audio(audio_file, format="audio/mp3")

if __name__ == "__main__":
    main()