File size: 2,779 Bytes
582aa2c 7d58c40 b698950 91c303a fa34db2 b698950 91c303a 7d58c40 91c303a 7632fc7 91c303a 7632fc7 7d58c40 91c303a 7632fc7 91c303a 7632fc7 91c303a abad73f 7d58c40 91c303a 7d58c40 91c303a |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 |
import streamlit as st
from diffusers import StableDiffusionPipeline
import torch
from PIL import Image
import tempfile
import whisper
import os
from realesrgan import RealESRGAN
from huggingface_hub import hf_hub_download
# Load Whisper for voice-to-text
@st.cache_resource
def load_whisper():
return whisper.load_model("base")
# Load Stable Diffusion models
@st.cache_resource
def load_pipelines():
pipelines = {
"Realistic": StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float32),
"Anime": StableDiffusionPipeline.from_pretrained("andite/anything-v4.0", torch_dtype=torch.float32),
"Ghibli": StableDiffusionPipeline.from_pretrained("nitrosocke/Ghibli-Diffusion", torch_dtype=torch.float32)
}
for key in pipelines:
pipelines[key] = pipelines[key].to("cpu")
return pipelines
# Load Real-ESRGAN for enhancement
@st.cache_resource
def load_enhancer():
from basicsr.archs.rrdbnet_arch import RRDBNet
model = RRDBNet(num_in_ch=3, num_out_ch=3, num_feat=64,
num_block=23, num_grow_ch=32, scale=4)
return RealESRGAN(device=torch.device("cpu"), scale=4).load_weights(model)
whisper_model = load_whisper()
pipes = load_pipelines()
enhancer = load_enhancer()
st.set_page_config(page_title="Love Text to Image", layout="centered")
st.title("💖 Love Text to Image Generator")
st.markdown("Write or speak a romantic poem and turn it into beautiful art ✨")
style = st.selectbox("Choose art style", ["Realistic", "Anime", "Ghibli"])
tab1, tab2 = st.tabs(["📝 Text Input", "🎤 Voice Input"])
prompt = ""
with tab1:
prompt = st.text_area("Enter your romantic poem or message")
with tab2:
audio_file = st.file_uploader("Upload your audio file (mp3, wav)", type=["mp3", "wav"])
if audio_file is not None:
with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp:
tmp.write(audio_file.read())
tmp_path = tmp.name
with st.spinner("Transcribing..."):
result = whisper_model.transcribe(tmp_path)
prompt = result["text"]
st.success("Transcription Complete!")
st.markdown(f"**Transcribed Text:** {prompt}")
os.remove(tmp_path)
if st.button("Generate Image"):
if prompt.strip() == "":
st.warning("Please enter or upload a love message.")
else:
with st.spinner("Generating image..."):
image = pipes[style](prompt).images[0]
st.image(image, caption="Original Generated Image", use_column_width=True)
with st.spinner("Enhancing image..."):
enhanced = enhancer.predict(image)
st.image(enhanced, caption="Enhanced Image", use_column_width=True)
|