app.py

import os
import torch
import gradio as gr
from PIL import Image
from diffusers import StableDiffusionPipeline
from openai import OpenAI
import speech_recognition as sr

# 配置模型与API密钥
OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
openai_client = OpenAI(api_key=OPENAI_API_KEY)
MODEL_ID = "runwayml/stable-diffusion-v1-5"

# 初始化 Stable Diffusion
device = "cpu"
pipe = StableDiffusionPipeline.from_pretrained(MODEL_ID, torch_dtype=torch.float32)
pipe = pipe.to(device)

# Prompt-to-Prompt 模块
def generate_prompt(description, provider, temperature):
    try:
        model = "gpt-3.5-turbo" if provider == "OpenAI" else "deepseek-chat"
        client = openai_client
        response = client.chat.completions.create(
            model=model,
            messages=[{"role": "user", "content": f"将以下描述转换为一个适用于 Stable Diffusion 的图像生成提示词：'{description}'"}],
            temperature=temperature
        )
        return response.choices[0].message.content.strip()
    except Exception as e:
        return f"❌ 提示生成失败：{str(e)}"

# Prompt-to-Image 模块
def generate_image(prompt, negative_prompt, guidance_scale, steps):
    with torch.no_grad():
        result = pipe(prompt=prompt, negative_prompt=negative_prompt, guidance_scale=guidance_scale, num_inference_steps=steps)
        return result.images[0]

# 语音转文字
def transcribe_audio(audio):
    recognizer = sr.Recognizer()
    with sr.AudioFile(audio) as source:
        audio_data = recognizer.record(source)
        try:
            return recognizer.recognize_google(audio_data, language="zh-CN")
        except sr.UnknownValueError:
            return "❌ 无法识别语音"
        except sr.RequestError as e:
            return f"❌ 语音识别出错: {e}"

# 构建 Gradio UI
with gr.Blocks(theme=gr.themes.Soft()) as demo:
    gr.Markdown("# 🎨 Prompt-to-Image 应用\n通过自然语言或语音描述生成图像")

    with gr.Row():
        with gr.Column():
            description = gr.Textbox(label="请输入图像描述", placeholder="如：空中的魔法树屋")
            audio_input = gr.Audio(type="filepath", label="录音或上传音频", sources=["microphone", "upload"])
            transcribed_text = gr.Textbox(label="语音转文字结果", interactive=False)
            transcribe_btn = gr.Button("识别语音")
            
            model_provider = gr.Dropdown(["OpenAI", "DeepSeek"], label="选择模型提供者")
            temperature = gr.Slider(0.1, 1.5, value=0.7, step=0.1, label="温度（Temperature）")
            generate_prompt_btn = gr.Button("生成提示词")

            generated_prompt = gr.Textbox(label="生成的提示词")
            negative_prompt = gr.Textbox(label="反向提示词", value="blurry, distorted, watermark")
            guidance = gr.Slider(1.0, 15.0, value=7.5, step=0.5, label="Guidance Scale")
            steps = gr.Slider(10, 50, value=25, step=1, label="Inference Steps")

            generate_image_btn = gr.Button("🎨 生成图像")

        with gr.Column():
            output_image = gr.Image(label="生成图像", type="pil")

    # 点击绑定
    transcribe_btn.click(fn=transcribe_audio, inputs=audio_input, outputs=transcribed_text)
    generate_prompt_btn.click(fn=generate_prompt, inputs=[description, model_provider, temperature], outputs=generated_prompt)
    generate_image_btn.click(fn=generate_image, inputs=[generated_prompt, negative_prompt, guidance, steps], outputs=output_image)

# 运行
if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860)