app.py

import streamlit as st
from transformers import pipeline
from gtts import gTTS
import io
import os

# function part
# img2text
def img2text(url):
    image_to_text_model = pipeline("image-to-text", model="Salesforce/blip-image-captioning-base")
    text = image_to_text_model(url)[0]["generated_text"]
    return text

# text2story
def text2story(text):
    story_generator = pipeline("text-generation", model="Qwen/QwQ-32B")
    story = story_generator(text, max_length=200, num_return_sequences=1)[0]["generated_text"]
    return story

# text2audio
def text2audio(story_text):
    # 创建 gTTS 对象
    tts = gTTS(text=story_text, lang='en')
    # 创建一个字节流对象用于存储音频数据
    audio_file = io.BytesIO()
    # 将音频数据写入字节流
    tts.write_to_fp(audio_file)
    # 将文件指针移到开头
    audio_file.seek(0)
    return audio_file

st.set_page_config(page_title="Your Image to Audio Story",
                   page_icon="🦜")
st.header("Turn Your Image to Audio Story")
uploaded_file = st.file_uploader("Select an Image...")

if uploaded_file is not None:
    # 保存上传的文件到临时文件
    temp_file_path = "temp_image.jpg"
    bytes_data = uploaded_file.getvalue()
    with open(temp_file_path, "wb") as file:
        file.write(bytes_data)
    st.image(uploaded_file, caption="Uploaded Image",
             use_column_width=True)

    # Stage 1: Image to Text
    st.text('Processing img2text...')
    scenario = img2text(temp_file_path)
    st.write(scenario)

    # 删除临时文件
    if os.path.exists(temp_file_path):
        os.remove(temp_file_path)

    # Stage 2: Text to Story
    st.text('Generating a story...')
    story = text2story(scenario)
    st.write(story)

    # Stage 3: Story to Audio data
    st.text('Generating audio data...')
    audio_data = text2audio(story)

    # Play button
    if st.button("Play Audio"):
        st.audio(audio_data,
                 format="audio/mpeg",
                 start_time=0)